test_transforms_v2_refactored.py 97.8 KB
Newer Older
1
import contextlib
2
import decimal
3
import inspect
Philip Meier's avatar
Philip Meier committed
4
import math
5
import pickle
6
import re
7
from pathlib import Path
8
9
10
11
12
13
14
15
16
17
18
19
20
from unittest import mock

import numpy as np
import PIL.Image
import pytest

import torch
import torchvision.transforms.v2 as transforms
from common_utils import (
    assert_equal,
    assert_no_warnings,
    cache,
    cpu_and_cuda,
21
    freeze_rng_state,
22
    ignore_jit_no_profile_information_warning,
23
    make_bounding_boxes,
24
25
    make_detection_mask,
    make_image,
26
27
    make_image_pil,
    make_image_tensor,
28
29
    make_segmentation_mask,
    make_video,
30
    make_video_tensor,
31
    needs_cuda,
Nicolas Hug's avatar
Nicolas Hug committed
32
    set_rng_seed,
33
)
34
35

from torch import nn
36
from torch.testing import assert_close
37
from torch.utils._pytree import tree_map
38
from torch.utils.data import DataLoader, default_collate
39
from torchvision import datapoints
Philip Meier's avatar
Philip Meier committed
40
41

from torchvision.transforms._functional_tensor import _max_value as get_max_value
42
43
from torchvision.transforms.functional import pil_modes_mapping
from torchvision.transforms.v2 import functional as F
44
from torchvision.transforms.v2.functional._utils import _get_kernel, _register_kernel_internal
45
46


Nicolas Hug's avatar
Nicolas Hug committed
47
48
49
50
51
52
@pytest.fixture(autouse=True)
def fix_rng_seed():
    set_rng_seed(0)
    yield


53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
def _to_tolerances(maybe_tolerance_dict):
    if not isinstance(maybe_tolerance_dict, dict):
        return dict(rtol=None, atol=None)

    tolerances = dict(rtol=0, atol=0)
    tolerances.update(maybe_tolerance_dict)
    return tolerances


def _check_kernel_cuda_vs_cpu(kernel, input, *args, rtol, atol, **kwargs):
    """Checks if the kernel produces closes results for inputs on GPU and CPU."""
    if input.device.type != "cuda":
        return

    input_cuda = input.as_subclass(torch.Tensor)
    input_cpu = input_cuda.to("cpu")

70
71
72
73
    with freeze_rng_state():
        actual = kernel(input_cuda, *args, **kwargs)
    with freeze_rng_state():
        expected = kernel(input_cpu, *args, **kwargs)
74
75
76
77
78

    assert_close(actual, expected, check_device=False, rtol=rtol, atol=atol)


@cache
79
def _script(obj):
80
    try:
81
        return torch.jit.script(obj)
82
    except Exception as error:
83
84
        name = getattr(obj, "__name__", obj.__class__.__name__)
        raise AssertionError(f"Trying to `torch.jit.script` '{name}' raised the error above.") from error
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140


def _check_kernel_scripted_vs_eager(kernel, input, *args, rtol, atol, **kwargs):
    """Checks if the kernel is scriptable and if the scripted output is close to the eager one."""
    if input.device.type != "cpu":
        return

    kernel_scripted = _script(kernel)

    input = input.as_subclass(torch.Tensor)
    with ignore_jit_no_profile_information_warning():
        actual = kernel_scripted(input, *args, **kwargs)
    expected = kernel(input, *args, **kwargs)

    assert_close(actual, expected, rtol=rtol, atol=atol)


def _check_kernel_batched_vs_unbatched(kernel, input, *args, rtol, atol, **kwargs):
    """Checks if the kernel produces close results for batched and unbatched inputs."""
    unbatched_input = input.as_subclass(torch.Tensor)

    for batch_dims in [(2,), (2, 1)]:
        repeats = [*batch_dims, *[1] * input.ndim]

        actual = kernel(unbatched_input.repeat(repeats), *args, **kwargs)

        expected = kernel(unbatched_input, *args, **kwargs)
        # We can't directly call `.repeat()` on the output, since some kernel also return some additional metadata
        if isinstance(expected, torch.Tensor):
            expected = expected.repeat(repeats)
        else:
            tensor, *metadata = expected
            expected = (tensor.repeat(repeats), *metadata)

        assert_close(actual, expected, rtol=rtol, atol=atol)

    for degenerate_batch_dims in [(0,), (5, 0), (0, 5)]:
        degenerate_batched_input = torch.empty(
            degenerate_batch_dims + input.shape, dtype=input.dtype, device=input.device
        )

        output = kernel(degenerate_batched_input, *args, **kwargs)
        # Most kernels just return a tensor, but some also return some additional metadata
        if not isinstance(output, torch.Tensor):
            output, *_ = output

        assert output.shape[: -input.ndim] == degenerate_batch_dims


def check_kernel(
    kernel,
    input,
    *args,
    check_cuda_vs_cpu=True,
    check_scripted_vs_eager=True,
    check_batched_vs_unbatched=True,
141
    expect_same_dtype=True,
142
143
144
145
146
147
148
149
150
151
152
153
    **kwargs,
):
    initial_input_version = input._version

    output = kernel(input.as_subclass(torch.Tensor), *args, **kwargs)
    # Most kernels just return a tensor, but some also return some additional metadata
    if not isinstance(output, torch.Tensor):
        output, *_ = output

    # check that no inplace operation happened
    assert input._version == initial_input_version

154
155
    if expect_same_dtype:
        assert output.dtype == input.dtype
156
157
158
159
160
161
162
163
164
165
166
167
    assert output.device == input.device

    if check_cuda_vs_cpu:
        _check_kernel_cuda_vs_cpu(kernel, input, *args, **kwargs, **_to_tolerances(check_cuda_vs_cpu))

    if check_scripted_vs_eager:
        _check_kernel_scripted_vs_eager(kernel, input, *args, **kwargs, **_to_tolerances(check_scripted_vs_eager))

    if check_batched_vs_unbatched:
        _check_kernel_batched_vs_unbatched(kernel, input, *args, **kwargs, **_to_tolerances(check_batched_vs_unbatched))


Nicolas Hug's avatar
Nicolas Hug committed
168
169
def _check_functional_scripted_smoke(functional, input, *args, **kwargs):
    """Checks if the functional can be scripted and the scripted version can be called without error."""
170
171
172
    if not isinstance(input, datapoints.Image):
        return

Nicolas Hug's avatar
Nicolas Hug committed
173
    functional_scripted = _script(functional)
174
    with ignore_jit_no_profile_information_warning():
Nicolas Hug's avatar
Nicolas Hug committed
175
        functional_scripted(input.as_subclass(torch.Tensor), *args, **kwargs)
176
177


Nicolas Hug's avatar
Nicolas Hug committed
178
def check_functional(functional, input, *args, check_scripted_smoke=True, **kwargs):
179
    unknown_input = object()
180
    with pytest.raises(TypeError, match=re.escape(str(type(unknown_input)))):
Nicolas Hug's avatar
Nicolas Hug committed
181
        functional(unknown_input, *args, **kwargs)
182

183
    with mock.patch("torch._C._log_api_usage_once", wraps=torch._C._log_api_usage_once) as spy:
Nicolas Hug's avatar
Nicolas Hug committed
184
        output = functional(input, *args, **kwargs)
185

Nicolas Hug's avatar
Nicolas Hug committed
186
        spy.assert_any_call(f"{functional.__module__}.{functional.__name__}")
187

188
189
190
191
192
    assert isinstance(output, type(input))

    if isinstance(input, datapoints.BoundingBoxes):
        assert output.format == input.format

193
    if check_scripted_smoke:
Nicolas Hug's avatar
Nicolas Hug committed
194
        _check_functional_scripted_smoke(functional, input, *args, **kwargs)
195
196


Nicolas Hug's avatar
Nicolas Hug committed
197
198
199
def check_functional_kernel_signature_match(functional, *, kernel, input_type):
    """Checks if the signature of the functional matches the kernel signature."""
    functional_params = list(inspect.signature(functional).parameters.values())[1:]
200
    kernel_params = list(inspect.signature(kernel).parameters.values())[1:]
201

202
    if issubclass(input_type, datapoints.Datapoint):
Nicolas Hug's avatar
Nicolas Hug committed
203
        # We filter out metadata that is implicitly passed to the functional through the input datapoint, but has to be
204
        # explicitly passed to the kernel.
205
206
207
208
        explicit_metadata = {
            datapoints.BoundingBoxes: {"format", "canvas_size"},
        }
        kernel_params = [param for param in kernel_params if param.name not in explicit_metadata.get(input_type, set())]
209

Nicolas Hug's avatar
Nicolas Hug committed
210
211
    functional_params = iter(functional_params)
    for functional_param, kernel_param in zip(functional_params, kernel_params):
212
        try:
Nicolas Hug's avatar
Nicolas Hug committed
213
214
215
216
            # In general, the functional parameters are a superset of the kernel parameters. Thus, we filter out
            # functional parameters that have no kernel equivalent while keeping the order intact.
            while functional_param.name != kernel_param.name:
                functional_param = next(functional_params)
217
218
219
        except StopIteration:
            raise AssertionError(
                f"Parameter `{kernel_param.name}` of kernel `{kernel.__name__}` "
Nicolas Hug's avatar
Nicolas Hug committed
220
                f"has no corresponding parameter on the functional `{functional.__name__}`."
221
222
223
224
225
            ) from None

        if issubclass(input_type, PIL.Image.Image):
            # PIL kernels often have more correct annotations, since they are not limited by JIT. Thus, we don't check
            # them in the first place.
Nicolas Hug's avatar
Nicolas Hug committed
226
            functional_param._annotation = kernel_param._annotation = inspect.Parameter.empty
227

Nicolas Hug's avatar
Nicolas Hug committed
228
        assert functional_param == kernel_param
229
230


231
def _check_transform_v1_compatibility(transform, input, rtol, atol):
232
    """If the transform defines the ``_v1_transform_cls`` attribute, checks if the transform has a public, static
233
234
235
    ``get_params`` method that is the v1 equivalent, the output is close to v1, is scriptable, and the scripted version
    can be called without error."""
    if type(input) is not torch.Tensor or isinstance(input, PIL.Image.Image):
236
237
        return

238
239
    v1_transform_cls = transform._v1_transform_cls
    if v1_transform_cls is None:
240
241
        return

242
243
    if hasattr(v1_transform_cls, "get_params"):
        assert type(transform).get_params is v1_transform_cls.get_params
244

245
246
247
248
249
250
251
252
253
    v1_transform = v1_transform_cls(**transform._extract_params_for_v1_transform())

    with freeze_rng_state():
        output_v2 = transform(input)

    with freeze_rng_state():
        output_v1 = v1_transform(input)

    assert_close(output_v2, output_v1, rtol=rtol, atol=atol)
254

255
256
257
258
    if isinstance(input, PIL.Image.Image):
        return

    _script(v1_transform)(input)
259
260


261
def check_transform(transform, input, check_v1_compatibility=True):
262
263
    pickle.loads(pickle.dumps(transform))

264
265
266
    output = transform(input)
    assert isinstance(output, type(input))

267
    if isinstance(input, datapoints.BoundingBoxes):
268
269
        assert output.format == input.format

270
271
    if check_v1_compatibility:
        _check_transform_v1_compatibility(transform, input, **_to_tolerances(check_v1_compatibility))
272
273


274
def transform_cls_to_functional(transform_cls, **transform_specific_kwargs):
275
    def wrapper(input, *args, **kwargs):
276
        transform = transform_cls(*args, **transform_specific_kwargs, **kwargs)
277
278
279
280
281
282
283
        return transform(input)

    wrapper.__name__ = transform_cls.__name__

    return wrapper


Philip Meier's avatar
Philip Meier committed
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
def param_value_parametrization(**kwargs):
    """Helper function to turn

    @pytest.mark.parametrize(
        ("param", "value"),
        ("a", 1),
        ("a", 2),
        ("a", 3),
        ("b", -1.0)
        ("b", 1.0)
    )

    into

    @param_value_parametrization(a=[1, 2, 3], b=[-1.0, 1.0])
    """
    return pytest.mark.parametrize(
        ("param", "value"),
        [(param, value) for param, values in kwargs.items() for value in values],
    )


def adapt_fill(value, *, dtype):
    """Adapt fill values in the range [0.0, 1.0] to the value range of the dtype"""
    if value is None:
        return value

    max_value = get_max_value(dtype)

    if isinstance(value, (int, float)):
        return type(value)(value * max_value)
    elif isinstance(value, (list, tuple)):
        return type(value)(type(v)(v * max_value) for v in value)
    else:
        raise ValueError(f"fill should be an int or float, or a list or tuple of the former, but got '{value}'.")


EXHAUSTIVE_TYPE_FILLS = [
    None,
    1,
    0.5,
    [1],
    [0.2],
    (0,),
    (0.7,),
    [1, 0, 1],
    [0.1, 0.2, 0.3],
    (0, 1, 0),
    (0.9, 0.234, 0.314),
]
CORRECTNESS_FILLS = [
    v for v in EXHAUSTIVE_TYPE_FILLS if v is None or isinstance(v, float) or (isinstance(v, list) and len(v) > 1)
]


339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
# We cannot use `list(transforms.InterpolationMode)` here, since it includes some PIL-only ones as well
INTERPOLATION_MODES = [
    transforms.InterpolationMode.NEAREST,
    transforms.InterpolationMode.NEAREST_EXACT,
    transforms.InterpolationMode.BILINEAR,
    transforms.InterpolationMode.BICUBIC,
]


@contextlib.contextmanager
def assert_warns_antialias_default_value():
    with pytest.warns(UserWarning, match="The default value of the antialias parameter of all the resizing transforms"):
        yield


Philip Meier's avatar
Philip Meier committed
354
def reference_affine_bounding_boxes_helper(bounding_boxes, *, format, canvas_size, affine_matrix):
355
    def transform(bbox):
356
357
358
359
        # Go to float before converting to prevent precision loss in case of CXCYWH -> XYXY and W or H is 1
        in_dtype = bbox.dtype
        if not torch.is_floating_point(bbox):
            bbox = bbox.float()
Nicolas Hug's avatar
Nicolas Hug committed
360
        bbox_xyxy = F.convert_bounding_box_format(
361
            bbox.as_subclass(torch.Tensor),
362
            old_format=format,
363
364
365
366
367
368
369
370
371
372
373
            new_format=datapoints.BoundingBoxFormat.XYXY,
            inplace=True,
        )
        points = np.array(
            [
                [bbox_xyxy[0].item(), bbox_xyxy[1].item(), 1.0],
                [bbox_xyxy[2].item(), bbox_xyxy[1].item(), 1.0],
                [bbox_xyxy[0].item(), bbox_xyxy[3].item(), 1.0],
                [bbox_xyxy[2].item(), bbox_xyxy[3].item(), 1.0],
            ]
        )
374
        transformed_points = np.matmul(points, affine_matrix.T)
375
376
377
378
379
380
381
382
383
        out_bbox = torch.tensor(
            [
                np.min(transformed_points[:, 0]).item(),
                np.min(transformed_points[:, 1]).item(),
                np.max(transformed_points[:, 0]).item(),
                np.max(transformed_points[:, 1]).item(),
            ],
            dtype=bbox_xyxy.dtype,
        )
Nicolas Hug's avatar
Nicolas Hug committed
384
        out_bbox = F.convert_bounding_box_format(
385
            out_bbox, old_format=datapoints.BoundingBoxFormat.XYXY, new_format=format, inplace=True
386
387
        )
        # It is important to clamp before casting, especially for CXCYWH format, dtype=int64
Philip Meier's avatar
Philip Meier committed
388
        out_bbox = F.clamp_bounding_boxes(out_bbox, format=format, canvas_size=canvas_size)
389
390
391
        out_bbox = out_bbox.to(dtype=in_dtype)
        return out_bbox

392
    return torch.stack([transform(b) for b in bounding_boxes.reshape(-1, 4).unbind()]).reshape(bounding_boxes.shape)
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454


class TestResize:
    INPUT_SIZE = (17, 11)
    OUTPUT_SIZES = [17, [17], (17,), [12, 13], (12, 13)]

    def _make_max_size_kwarg(self, *, use_max_size, size):
        if use_max_size:
            if not (isinstance(size, int) or len(size) == 1):
                # This would result in an `ValueError`
                return None

            max_size = (size if isinstance(size, int) else size[0]) + 1
        else:
            max_size = None

        return dict(max_size=max_size)

    def _compute_output_size(self, *, input_size, size, max_size):
        if not (isinstance(size, int) or len(size) == 1):
            return tuple(size)

        if not isinstance(size, int):
            size = size[0]

        old_height, old_width = input_size
        ratio = old_width / old_height
        if ratio > 1:
            new_height = size
            new_width = int(ratio * new_height)
        else:
            new_width = size
            new_height = int(new_width / ratio)

        if max_size is not None and max(new_height, new_width) > max_size:
            # Need to recompute the aspect ratio, since it might have changed due to rounding
            ratio = new_width / new_height
            if ratio > 1:
                new_width = max_size
                new_height = int(new_width / ratio)
            else:
                new_height = max_size
                new_width = int(new_height * ratio)

        return new_height, new_width

    @pytest.mark.parametrize("size", OUTPUT_SIZES)
    @pytest.mark.parametrize("interpolation", INTERPOLATION_MODES)
    @pytest.mark.parametrize("use_max_size", [True, False])
    @pytest.mark.parametrize("antialias", [True, False])
    @pytest.mark.parametrize("dtype", [torch.float32, torch.uint8])
    @pytest.mark.parametrize("device", cpu_and_cuda())
    def test_kernel_image_tensor(self, size, interpolation, use_max_size, antialias, dtype, device):
        if not (max_size_kwarg := self._make_max_size_kwarg(use_max_size=use_max_size, size=size)):
            return

        # In contrast to CPU, there is no native `InterpolationMode.BICUBIC` implementation for uint8 images on CUDA.
        # Internally, it uses the float path. Thus, we need to test with an enormous tolerance here to account for that.
        atol = 30 if transforms.InterpolationMode.BICUBIC and dtype is torch.uint8 else 1
        check_cuda_vs_cpu_tolerances = dict(rtol=0, atol=atol / 255 if dtype.is_floating_point else atol)

        check_kernel(
455
            F.resize_image,
456
            make_image(self.INPUT_SIZE, dtype=dtype, device=device),
457
458
459
460
461
462
463
464
465
466
467
468
469
            size=size,
            interpolation=interpolation,
            **max_size_kwarg,
            antialias=antialias,
            check_cuda_vs_cpu=check_cuda_vs_cpu_tolerances,
            check_scripted_vs_eager=not isinstance(size, int),
        )

    @pytest.mark.parametrize("format", list(datapoints.BoundingBoxFormat))
    @pytest.mark.parametrize("size", OUTPUT_SIZES)
    @pytest.mark.parametrize("use_max_size", [True, False])
    @pytest.mark.parametrize("dtype", [torch.float32, torch.int64])
    @pytest.mark.parametrize("device", cpu_and_cuda())
470
    def test_kernel_bounding_boxes(self, format, size, use_max_size, dtype, device):
471
472
473
        if not (max_size_kwarg := self._make_max_size_kwarg(use_max_size=use_max_size, size=size)):
            return

474
        bounding_boxes = make_bounding_boxes(
475
            format=format,
Philip Meier's avatar
Philip Meier committed
476
            canvas_size=self.INPUT_SIZE,
477
478
            dtype=dtype,
            device=device,
Philip Meier's avatar
Philip Meier committed
479
        )
480
        check_kernel(
481
482
            F.resize_bounding_boxes,
            bounding_boxes,
Philip Meier's avatar
Philip Meier committed
483
            canvas_size=bounding_boxes.canvas_size,
484
485
486
487
488
            size=size,
            **max_size_kwarg,
            check_scripted_vs_eager=not isinstance(size, int),
        )

489
490
491
    @pytest.mark.parametrize("make_mask", [make_segmentation_mask, make_detection_mask])
    def test_kernel_mask(self, make_mask):
        check_kernel(F.resize_mask, make_mask(self.INPUT_SIZE), size=self.OUTPUT_SIZES[-1])
492
493

    def test_kernel_video(self):
494
        check_kernel(F.resize_video, make_video(self.INPUT_SIZE), size=self.OUTPUT_SIZES[-1], antialias=True)
495
496
497

    @pytest.mark.parametrize("size", OUTPUT_SIZES)
    @pytest.mark.parametrize(
Philip Meier's avatar
Philip Meier committed
498
        "make_input",
499
        [make_image_tensor, make_image_pil, make_image, make_bounding_boxes, make_segmentation_mask, make_video],
500
    )
Nicolas Hug's avatar
Nicolas Hug committed
501
502
    def test_functional(self, size, make_input):
        check_functional(
503
            F.resize,
504
            make_input(self.INPUT_SIZE),
505
506
507
508
509
510
            size=size,
            antialias=True,
            check_scripted_smoke=not isinstance(size, int),
        )

    @pytest.mark.parametrize(
511
        ("kernel", "input_type"),
512
        [
513
514
515
            (F.resize_image, torch.Tensor),
            (F._resize_image_pil, PIL.Image.Image),
            (F.resize_image, datapoints.Image),
516
            (F.resize_bounding_boxes, datapoints.BoundingBoxes),
517
518
            (F.resize_mask, datapoints.Mask),
            (F.resize_video, datapoints.Video),
519
520
        ],
    )
Nicolas Hug's avatar
Nicolas Hug committed
521
522
    def test_functional_signature(self, kernel, input_type):
        check_functional_kernel_signature_match(F.resize, kernel=kernel, input_type=input_type)
523
524
525
526

    @pytest.mark.parametrize("size", OUTPUT_SIZES)
    @pytest.mark.parametrize("device", cpu_and_cuda())
    @pytest.mark.parametrize(
527
528
529
530
531
        "make_input",
        [
            make_image_tensor,
            make_image_pil,
            make_image,
532
            make_bounding_boxes,
533
534
535
536
            make_segmentation_mask,
            make_detection_mask,
            make_video,
        ],
537
    )
538
    def test_transform(self, size, device, make_input):
539
540
541
542
543
544
        check_transform(
            transforms.Resize(size=size, antialias=True),
            make_input(self.INPUT_SIZE, device=device),
            # atol=1 due to Resize v2 is using native uint8 interpolate path for bilinear and nearest modes
            check_v1_compatibility=dict(rtol=0, atol=1),
        )
545
546

    def _check_output_size(self, input, output, *, size, max_size):
Philip Meier's avatar
Philip Meier committed
547
548
        assert tuple(F.get_size(output)) == self._compute_output_size(
            input_size=F.get_size(input), size=size, max_size=max_size
549
550
551
552
553
554
555
556
557
558
559
560
        )

    @pytest.mark.parametrize("size", OUTPUT_SIZES)
    # `InterpolationMode.NEAREST` is modeled after the buggy `INTER_NEAREST` interpolation of CV2.
    # The PIL equivalent of `InterpolationMode.NEAREST` is `InterpolationMode.NEAREST_EXACT`
    @pytest.mark.parametrize("interpolation", set(INTERPOLATION_MODES) - {transforms.InterpolationMode.NEAREST})
    @pytest.mark.parametrize("use_max_size", [True, False])
    @pytest.mark.parametrize("fn", [F.resize, transform_cls_to_functional(transforms.Resize)])
    def test_image_correctness(self, size, interpolation, use_max_size, fn):
        if not (max_size_kwarg := self._make_max_size_kwarg(use_max_size=use_max_size, size=size)):
            return

561
        image = make_image(self.INPUT_SIZE, dtype=torch.uint8)
562
563

        actual = fn(image, size=size, interpolation=interpolation, **max_size_kwarg, antialias=True)
564
        expected = F.to_image(F.resize(F.to_pil_image(image), size=size, interpolation=interpolation, **max_size_kwarg))
565
566
567
568

        self._check_output_size(image, actual, size=size, **max_size_kwarg)
        torch.testing.assert_close(actual, expected, atol=1, rtol=0)

569
    def _reference_resize_bounding_boxes(self, bounding_boxes, *, size, max_size=None):
Philip Meier's avatar
Philip Meier committed
570
        old_height, old_width = bounding_boxes.canvas_size
571
        new_height, new_width = self._compute_output_size(
Philip Meier's avatar
Philip Meier committed
572
            input_size=bounding_boxes.canvas_size, size=size, max_size=max_size
573
574
575
        )

        if (old_height, old_width) == (new_height, new_width):
576
            return bounding_boxes
577
578
579
580
581
582

        affine_matrix = np.array(
            [
                [new_width / old_width, 0, 0],
                [0, new_height / old_height, 0],
            ],
583
            dtype="float64" if bounding_boxes.dtype == torch.float64 else "float32",
584
585
        )

586
587
588
        expected_bboxes = reference_affine_bounding_boxes_helper(
            bounding_boxes,
            format=bounding_boxes.format,
Philip Meier's avatar
Philip Meier committed
589
            canvas_size=(new_height, new_width),
590
591
            affine_matrix=affine_matrix,
        )
592
        return datapoints.wrap(expected_bboxes, like=bounding_boxes, canvas_size=(new_height, new_width))
593
594
595
596
597

    @pytest.mark.parametrize("format", list(datapoints.BoundingBoxFormat))
    @pytest.mark.parametrize("size", OUTPUT_SIZES)
    @pytest.mark.parametrize("use_max_size", [True, False])
    @pytest.mark.parametrize("fn", [F.resize, transform_cls_to_functional(transforms.Resize)])
598
    def test_bounding_boxes_correctness(self, format, size, use_max_size, fn):
599
600
601
        if not (max_size_kwarg := self._make_max_size_kwarg(use_max_size=use_max_size, size=size)):
            return

602
        bounding_boxes = make_bounding_boxes(format=format, canvas_size=self.INPUT_SIZE)
603

604
605
        actual = fn(bounding_boxes, size=size, **max_size_kwarg)
        expected = self._reference_resize_bounding_boxes(bounding_boxes, size=size, **max_size_kwarg)
606

607
        self._check_output_size(bounding_boxes, actual, size=size, **max_size_kwarg)
608
609
610
611
        torch.testing.assert_close(actual, expected)

    @pytest.mark.parametrize("interpolation", set(transforms.InterpolationMode) - set(INTERPOLATION_MODES))
    @pytest.mark.parametrize(
612
613
        "make_input",
        [make_image_tensor, make_image_pil, make_image, make_video],
614
    )
615
616
    def test_pil_interpolation_compat_smoke(self, interpolation, make_input):
        input = make_input(self.INPUT_SIZE)
617
618
619
620
621
622
623
624
625
626
627
628
629

        with (
            contextlib.nullcontext()
            if isinstance(input, PIL.Image.Image)
            # This error is triggered in PyTorch core
            else pytest.raises(NotImplementedError, match=f"got {interpolation.value.lower()}")
        ):
            F.resize(
                input,
                size=self.OUTPUT_SIZES[0],
                interpolation=interpolation,
            )

Nicolas Hug's avatar
Nicolas Hug committed
630
    def test_functional_pil_antialias_warning(self):
631
        with pytest.warns(UserWarning, match="Anti-alias option is always applied for PIL Image input"):
632
            F.resize(make_image_pil(self.INPUT_SIZE), size=self.OUTPUT_SIZES[0], antialias=False)
633
634
635

    @pytest.mark.parametrize("size", OUTPUT_SIZES)
    @pytest.mark.parametrize(
636
637
638
639
640
        "make_input",
        [
            make_image_tensor,
            make_image_pil,
            make_image,
641
            make_bounding_boxes,
642
643
644
645
            make_segmentation_mask,
            make_detection_mask,
            make_video,
        ],
646
    )
647
    def test_max_size_error(self, size, make_input):
648
649
650
651
652
653
654
655
656
        if isinstance(size, int) or len(size) == 1:
            max_size = (size if isinstance(size, int) else size[0]) - 1
            match = "must be strictly greater than the requested size"
        else:
            # value can be anything other than None
            max_size = -1
            match = "size should be an int or a sequence of length 1"

        with pytest.raises(ValueError, match=match):
657
            F.resize(make_input(self.INPUT_SIZE), size=size, max_size=max_size, antialias=True)
658
659
660

    @pytest.mark.parametrize("interpolation", INTERPOLATION_MODES)
    @pytest.mark.parametrize(
661
662
        "make_input",
        [make_image_tensor, make_image, make_video],
663
    )
664
    def test_antialias_warning(self, interpolation, make_input):
665
666
667
668
669
        with (
            assert_warns_antialias_default_value()
            if interpolation in {transforms.InterpolationMode.BILINEAR, transforms.InterpolationMode.BICUBIC}
            else assert_no_warnings()
        ):
Philip Meier's avatar
Philip Meier committed
670
            F.resize(
671
                make_input(self.INPUT_SIZE),
Philip Meier's avatar
Philip Meier committed
672
673
674
                size=self.OUTPUT_SIZES[0],
                interpolation=interpolation,
            )
675
676
677

    @pytest.mark.parametrize("interpolation", INTERPOLATION_MODES)
    @pytest.mark.parametrize(
678
679
        "make_input",
        [make_image_tensor, make_image_pil, make_image, make_video],
680
    )
681
682
683
    def test_interpolation_int(self, interpolation, make_input):
        input = make_input(self.INPUT_SIZE)

684
685
686
        # `InterpolationMode.NEAREST_EXACT` has no proper corresponding integer equivalent. Internally, we map it to
        # `0` to be the same as `InterpolationMode.NEAREST` for PIL. However, for the tensor backend there is a
        # difference and thus we don't test it here.
687
        if isinstance(input, torch.Tensor) and interpolation is transforms.InterpolationMode.NEAREST_EXACT:
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
            return

        expected = F.resize(input, size=self.OUTPUT_SIZES[0], interpolation=interpolation, antialias=True)
        actual = F.resize(
            input, size=self.OUTPUT_SIZES[0], interpolation=pil_modes_mapping[interpolation], antialias=True
        )

        assert_equal(actual, expected)

    def test_transform_unknown_size_error(self):
        with pytest.raises(ValueError, match="size can either be an integer or a list or tuple of one or two integers"):
            transforms.Resize(size=object())

    @pytest.mark.parametrize(
        "size", [min(INPUT_SIZE), [min(INPUT_SIZE)], (min(INPUT_SIZE),), list(INPUT_SIZE), tuple(INPUT_SIZE)]
    )
    @pytest.mark.parametrize(
705
706
707
708
709
        "make_input",
        [
            make_image_tensor,
            make_image_pil,
            make_image,
710
            make_bounding_boxes,
711
712
713
714
            make_segmentation_mask,
            make_detection_mask,
            make_video,
        ],
715
    )
716
717
    def test_noop(self, size, make_input):
        input = make_input(self.INPUT_SIZE)
718

Philip Meier's avatar
Philip Meier committed
719
        output = F.resize(input, size=F.get_size(input), antialias=True)
720
721
722

        # This identity check is not a requirement. It is here to avoid breaking the behavior by accident. If there
        # is a good reason to break this, feel free to downgrade to an equality check.
723
        if isinstance(input, datapoints.Datapoint):
724
725
726
727
728
729
730
731
            # We can't test identity directly, since that checks for the identity of the Python object. Since all
            # datapoints unwrap before a kernel and wrap again afterwards, the Python object changes. Thus, we check
            # that the underlying storage is the same
            assert output.data_ptr() == input.data_ptr()
        else:
            assert output is input

    @pytest.mark.parametrize(
732
733
734
735
736
        "make_input",
        [
            make_image_tensor,
            make_image_pil,
            make_image,
737
            make_bounding_boxes,
738
739
740
741
            make_segmentation_mask,
            make_detection_mask,
            make_video,
        ],
742
    )
743
    def test_no_regression_5405(self, make_input):
744
745
746
        # Checks that `max_size` is not ignored if `size == small_edge_size`
        # See https://github.com/pytorch/vision/issues/5405

747
        input = make_input(self.INPUT_SIZE)
748

Philip Meier's avatar
Philip Meier committed
749
        size = min(F.get_size(input))
750
751
752
        max_size = size + 1
        output = F.resize(input, size=size, max_size=max_size, antialias=True)

Philip Meier's avatar
Philip Meier committed
753
        assert max(F.get_size(output)) == max_size
754

755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
    def _make_image(self, *args, batch_dims=(), memory_format=torch.contiguous_format, **kwargs):
        # torch.channels_last memory_format is only available for 4D tensors, i.e. (B, C, H, W). However, images coming
        # from PIL or our own I/O functions do not have a batch dimensions and are thus 3D, i.e. (C, H, W). Still, the
        # layout of the data in memory is channels last. To emulate this when a 3D input is requested here, we create
        # the image as 4D and create a view with the right shape afterwards. With this the layout in memory is channels
        # last although PyTorch doesn't recognizes it as such.
        emulate_channels_last = memory_format is torch.channels_last and len(batch_dims) != 1

        image = make_image(
            *args,
            batch_dims=(math.prod(batch_dims),) if emulate_channels_last else batch_dims,
            memory_format=memory_format,
            **kwargs,
        )

        if emulate_channels_last:
            image = datapoints.wrap(image.view(*batch_dims, *image.shape[-3:]), like=image)

        return image

    def _check_stride(self, image, *, memory_format):
        C, H, W = F.get_dimensions(image)
        if memory_format is torch.contiguous_format:
            expected_stride = (H * W, W, 1)
        elif memory_format is torch.channels_last:
            expected_stride = (1, W * C, C)
        else:
            raise ValueError(f"Unknown memory_format: {memory_format}")

        assert image.stride() == expected_stride

    # TODO: We can remove this test and related torchvision workaround
    #  once we fixed related pytorch issue: https://github.com/pytorch/pytorch/issues/68430
    @pytest.mark.parametrize("interpolation", INTERPOLATION_MODES)
    @pytest.mark.parametrize("antialias", [True, False])
    @pytest.mark.parametrize("memory_format", [torch.contiguous_format, torch.channels_last])
    @pytest.mark.parametrize("dtype", [torch.uint8, torch.float32])
    @pytest.mark.parametrize("device", cpu_and_cuda())
    def test_kernel_image_memory_format_consistency(self, interpolation, antialias, memory_format, dtype, device):
        size = self.OUTPUT_SIZES[0]

        input = self._make_image(self.INPUT_SIZE, dtype=dtype, device=device, memory_format=memory_format)

        # Smoke test to make sure we aren't starting with wrong assumptions
        self._check_stride(input, memory_format=memory_format)

        output = F.resize_image(input, size=size, interpolation=interpolation, antialias=antialias)

        self._check_stride(output, memory_format=memory_format)

    def test_float16_no_rounding(self):
        # Make sure Resize() doesn't round float16 images
        # Non-regression test for https://github.com/pytorch/vision/issues/7667

        input = make_image_tensor(self.INPUT_SIZE, dtype=torch.float16)
        output = F.resize_image(input, size=self.OUTPUT_SIZES[0])

        assert output.dtype is torch.float16
        assert (output.round() - output).abs().sum() > 0

815
816
817
818
819

class TestHorizontalFlip:
    @pytest.mark.parametrize("dtype", [torch.float32, torch.uint8])
    @pytest.mark.parametrize("device", cpu_and_cuda())
    def test_kernel_image_tensor(self, dtype, device):
820
        check_kernel(F.horizontal_flip_image, make_image(dtype=dtype, device=device))
821
822
823
824

    @pytest.mark.parametrize("format", list(datapoints.BoundingBoxFormat))
    @pytest.mark.parametrize("dtype", [torch.float32, torch.int64])
    @pytest.mark.parametrize("device", cpu_and_cuda())
825
    def test_kernel_bounding_boxes(self, format, dtype, device):
826
        bounding_boxes = make_bounding_boxes(format=format, dtype=dtype, device=device)
827
        check_kernel(
828
829
            F.horizontal_flip_bounding_boxes,
            bounding_boxes,
830
            format=format,
Philip Meier's avatar
Philip Meier committed
831
            canvas_size=bounding_boxes.canvas_size,
832
833
        )

834
835
836
    @pytest.mark.parametrize("make_mask", [make_segmentation_mask, make_detection_mask])
    def test_kernel_mask(self, make_mask):
        check_kernel(F.horizontal_flip_mask, make_mask())
837
838

    def test_kernel_video(self):
839
        check_kernel(F.horizontal_flip_video, make_video())
840
841

    @pytest.mark.parametrize(
Philip Meier's avatar
Philip Meier committed
842
        "make_input",
843
        [make_image_tensor, make_image_pil, make_image, make_bounding_boxes, make_segmentation_mask, make_video],
844
    )
Nicolas Hug's avatar
Nicolas Hug committed
845
846
    def test_functional(self, make_input):
        check_functional(F.horizontal_flip, make_input())
847
848

    @pytest.mark.parametrize(
849
        ("kernel", "input_type"),
850
        [
851
852
853
            (F.horizontal_flip_image, torch.Tensor),
            (F._horizontal_flip_image_pil, PIL.Image.Image),
            (F.horizontal_flip_image, datapoints.Image),
854
            (F.horizontal_flip_bounding_boxes, datapoints.BoundingBoxes),
855
856
            (F.horizontal_flip_mask, datapoints.Mask),
            (F.horizontal_flip_video, datapoints.Video),
857
858
        ],
    )
Nicolas Hug's avatar
Nicolas Hug committed
859
860
    def test_functional_signature(self, kernel, input_type):
        check_functional_kernel_signature_match(F.horizontal_flip, kernel=kernel, input_type=input_type)
861
862

    @pytest.mark.parametrize(
863
        "make_input",
864
        [make_image_tensor, make_image_pil, make_image, make_bounding_boxes, make_segmentation_mask, make_video],
865
866
    )
    @pytest.mark.parametrize("device", cpu_and_cuda())
867
    def test_transform(self, make_input, device):
868
        check_transform(transforms.RandomHorizontalFlip(p=1), make_input(device=device))
869
870
871
872
873

    @pytest.mark.parametrize(
        "fn", [F.horizontal_flip, transform_cls_to_functional(transforms.RandomHorizontalFlip, p=1)]
    )
    def test_image_correctness(self, fn):
874
        image = make_image(dtype=torch.uint8, device="cpu")
875
876

        actual = fn(image)
877
        expected = F.to_image(F.horizontal_flip(F.to_pil_image(image)))
878
879
880

        torch.testing.assert_close(actual, expected)

881
    def _reference_horizontal_flip_bounding_boxes(self, bounding_boxes):
882
883
        affine_matrix = np.array(
            [
Philip Meier's avatar
Philip Meier committed
884
                [-1, 0, bounding_boxes.canvas_size[1]],
885
886
                [0, 1, 0],
            ],
887
            dtype="float64" if bounding_boxes.dtype == torch.float64 else "float32",
888
889
        )

890
891
892
        expected_bboxes = reference_affine_bounding_boxes_helper(
            bounding_boxes,
            format=bounding_boxes.format,
Philip Meier's avatar
Philip Meier committed
893
            canvas_size=bounding_boxes.canvas_size,
894
895
896
            affine_matrix=affine_matrix,
        )

897
        return datapoints.wrap(expected_bboxes, like=bounding_boxes)
898
899
900
901
902

    @pytest.mark.parametrize("format", list(datapoints.BoundingBoxFormat))
    @pytest.mark.parametrize(
        "fn", [F.horizontal_flip, transform_cls_to_functional(transforms.RandomHorizontalFlip, p=1)]
    )
903
    def test_bounding_boxes_correctness(self, format, fn):
904
        bounding_boxes = make_bounding_boxes(format=format)
905

906
907
        actual = fn(bounding_boxes)
        expected = self._reference_horizontal_flip_bounding_boxes(bounding_boxes)
908
909
910
911

        torch.testing.assert_close(actual, expected)

    @pytest.mark.parametrize(
912
        "make_input",
913
        [make_image_tensor, make_image_pil, make_image, make_bounding_boxes, make_segmentation_mask, make_video],
914
915
    )
    @pytest.mark.parametrize("device", cpu_and_cuda())
916
917
    def test_transform_noop(self, make_input, device):
        input = make_input(device=device)
918
919
920
921
922
923

        transform = transforms.RandomHorizontalFlip(p=0)

        output = transform(input)

        assert_equal(output, input)
Philip Meier's avatar
Philip Meier committed
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966


class TestAffine:
    _EXHAUSTIVE_TYPE_AFFINE_KWARGS = dict(
        # float, int
        angle=[-10.9, 18],
        # two-list of float, two-list of int, two-tuple of float, two-tuple of int
        translate=[[6.3, -0.6], [1, -3], (16.6, -6.6), (-2, 4)],
        # float
        scale=[0.5],
        # float, int,
        # one-list of float, one-list of int, one-tuple of float, one-tuple of int
        # two-list of float, two-list of int, two-tuple of float, two-tuple of int
        shear=[35.6, 38, [-37.7], [-23], (5.3,), (-52,), [5.4, 21.8], [-47, 51], (-11.2, 36.7), (8, -53)],
        # None
        # two-list of float, two-list of int, two-tuple of float, two-tuple of int
        center=[None, [1.2, 4.9], [-3, 1], (2.5, -4.7), (3, 2)],
    )
    # The special case for shear makes sure we pick a value that is supported while JIT scripting
    _MINIMAL_AFFINE_KWARGS = {
        k: vs[0] if k != "shear" else next(v for v in vs if isinstance(v, list))
        for k, vs in _EXHAUSTIVE_TYPE_AFFINE_KWARGS.items()
    }
    _CORRECTNESS_AFFINE_KWARGS = {
        k: [v for v in vs if v is None or isinstance(v, float) or (isinstance(v, list) and len(v) > 1)]
        for k, vs in _EXHAUSTIVE_TYPE_AFFINE_KWARGS.items()
    }

    _EXHAUSTIVE_TYPE_TRANSFORM_AFFINE_RANGES = dict(
        degrees=[30, (-15, 20)],
        translate=[None, (0.5, 0.5)],
        scale=[None, (0.75, 1.25)],
        shear=[None, (12, 30, -17, 5), 10, (-5, 12)],
    )
    _CORRECTNESS_TRANSFORM_AFFINE_RANGES = {
        k: next(v for v in vs if v is not None) for k, vs in _EXHAUSTIVE_TYPE_TRANSFORM_AFFINE_RANGES.items()
    }

    def _check_kernel(self, kernel, input, *args, **kwargs):
        kwargs_ = self._MINIMAL_AFFINE_KWARGS.copy()
        kwargs_.update(kwargs)
        check_kernel(kernel, input, *args, **kwargs_)

Philip Meier's avatar
Philip Meier committed
967
968
969
970
971
972
973
    @param_value_parametrization(
        angle=_EXHAUSTIVE_TYPE_AFFINE_KWARGS["angle"],
        translate=_EXHAUSTIVE_TYPE_AFFINE_KWARGS["translate"],
        shear=_EXHAUSTIVE_TYPE_AFFINE_KWARGS["shear"],
        center=_EXHAUSTIVE_TYPE_AFFINE_KWARGS["center"],
        interpolation=[transforms.InterpolationMode.NEAREST, transforms.InterpolationMode.BILINEAR],
        fill=EXHAUSTIVE_TYPE_FILLS,
Philip Meier's avatar
Philip Meier committed
974
975
976
977
978
    )
    @pytest.mark.parametrize("dtype", [torch.float32, torch.uint8])
    @pytest.mark.parametrize("device", cpu_and_cuda())
    def test_kernel_image_tensor(self, param, value, dtype, device):
        if param == "fill":
Philip Meier's avatar
Philip Meier committed
979
            value = adapt_fill(value, dtype=dtype)
Philip Meier's avatar
Philip Meier committed
980
        self._check_kernel(
981
            F.affine_image,
982
            make_image(dtype=dtype, device=device),
Philip Meier's avatar
Philip Meier committed
983
984
985
986
987
988
989
            **{param: value},
            check_scripted_vs_eager=not (param in {"shear", "fill"} and isinstance(value, (int, float))),
            check_cuda_vs_cpu=dict(atol=1, rtol=0)
            if dtype is torch.uint8 and param == "interpolation" and value is transforms.InterpolationMode.BILINEAR
            else True,
        )

Philip Meier's avatar
Philip Meier committed
990
991
992
993
994
    @param_value_parametrization(
        angle=_EXHAUSTIVE_TYPE_AFFINE_KWARGS["angle"],
        translate=_EXHAUSTIVE_TYPE_AFFINE_KWARGS["translate"],
        shear=_EXHAUSTIVE_TYPE_AFFINE_KWARGS["shear"],
        center=_EXHAUSTIVE_TYPE_AFFINE_KWARGS["center"],
Philip Meier's avatar
Philip Meier committed
995
996
997
998
    )
    @pytest.mark.parametrize("format", list(datapoints.BoundingBoxFormat))
    @pytest.mark.parametrize("dtype", [torch.float32, torch.int64])
    @pytest.mark.parametrize("device", cpu_and_cuda())
999
    def test_kernel_bounding_boxes(self, param, value, format, dtype, device):
1000
        bounding_boxes = make_bounding_boxes(format=format, dtype=dtype, device=device)
Philip Meier's avatar
Philip Meier committed
1001
        self._check_kernel(
1002
1003
            F.affine_bounding_boxes,
            bounding_boxes,
Philip Meier's avatar
Philip Meier committed
1004
            format=format,
Philip Meier's avatar
Philip Meier committed
1005
            canvas_size=bounding_boxes.canvas_size,
Philip Meier's avatar
Philip Meier committed
1006
1007
1008
1009
            **{param: value},
            check_scripted_vs_eager=not (param == "shear" and isinstance(value, (int, float))),
        )

1010
1011
1012
    @pytest.mark.parametrize("make_mask", [make_segmentation_mask, make_detection_mask])
    def test_kernel_mask(self, make_mask):
        self._check_kernel(F.affine_mask, make_mask())
Philip Meier's avatar
Philip Meier committed
1013
1014

    def test_kernel_video(self):
1015
        self._check_kernel(F.affine_video, make_video())
Philip Meier's avatar
Philip Meier committed
1016
1017

    @pytest.mark.parametrize(
Philip Meier's avatar
Philip Meier committed
1018
        "make_input",
1019
        [make_image_tensor, make_image_pil, make_image, make_bounding_boxes, make_segmentation_mask, make_video],
Philip Meier's avatar
Philip Meier committed
1020
    )
Nicolas Hug's avatar
Nicolas Hug committed
1021
1022
    def test_functional(self, make_input):
        check_functional(F.affine, make_input(), **self._MINIMAL_AFFINE_KWARGS)
Philip Meier's avatar
Philip Meier committed
1023
1024

    @pytest.mark.parametrize(
1025
        ("kernel", "input_type"),
Philip Meier's avatar
Philip Meier committed
1026
        [
1027
1028
1029
            (F.affine_image, torch.Tensor),
            (F._affine_image_pil, PIL.Image.Image),
            (F.affine_image, datapoints.Image),
1030
            (F.affine_bounding_boxes, datapoints.BoundingBoxes),
1031
1032
            (F.affine_mask, datapoints.Mask),
            (F.affine_video, datapoints.Video),
Philip Meier's avatar
Philip Meier committed
1033
1034
        ],
    )
Nicolas Hug's avatar
Nicolas Hug committed
1035
1036
    def test_functional_signature(self, kernel, input_type):
        check_functional_kernel_signature_match(F.affine, kernel=kernel, input_type=input_type)
Philip Meier's avatar
Philip Meier committed
1037
1038

    @pytest.mark.parametrize(
1039
        "make_input",
1040
        [make_image_tensor, make_image_pil, make_image, make_bounding_boxes, make_segmentation_mask, make_video],
Philip Meier's avatar
Philip Meier committed
1041
1042
    )
    @pytest.mark.parametrize("device", cpu_and_cuda())
1043
1044
    def test_transform(self, make_input, device):
        input = make_input(device=device)
Philip Meier's avatar
Philip Meier committed
1045

1046
        check_transform(transforms.RandomAffine(**self._CORRECTNESS_TRANSFORM_AFFINE_RANGES), input)
Philip Meier's avatar
Philip Meier committed
1047
1048
1049
1050
1051
1052
1053
1054
1055

    @pytest.mark.parametrize("angle", _CORRECTNESS_AFFINE_KWARGS["angle"])
    @pytest.mark.parametrize("translate", _CORRECTNESS_AFFINE_KWARGS["translate"])
    @pytest.mark.parametrize("scale", _CORRECTNESS_AFFINE_KWARGS["scale"])
    @pytest.mark.parametrize("shear", _CORRECTNESS_AFFINE_KWARGS["shear"])
    @pytest.mark.parametrize("center", _CORRECTNESS_AFFINE_KWARGS["center"])
    @pytest.mark.parametrize(
        "interpolation", [transforms.InterpolationMode.NEAREST, transforms.InterpolationMode.BILINEAR]
    )
Philip Meier's avatar
Philip Meier committed
1056
    @pytest.mark.parametrize("fill", CORRECTNESS_FILLS)
Philip Meier's avatar
Philip Meier committed
1057
    def test_functional_image_correctness(self, angle, translate, scale, shear, center, interpolation, fill):
1058
        image = make_image(dtype=torch.uint8, device="cpu")
Philip Meier's avatar
Philip Meier committed
1059

Philip Meier's avatar
Philip Meier committed
1060
        fill = adapt_fill(fill, dtype=torch.uint8)
Philip Meier's avatar
Philip Meier committed
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071

        actual = F.affine(
            image,
            angle=angle,
            translate=translate,
            scale=scale,
            shear=shear,
            center=center,
            interpolation=interpolation,
            fill=fill,
        )
1072
        expected = F.to_image(
Philip Meier's avatar
Philip Meier committed
1073
            F.affine(
1074
                F.to_pil_image(image),
Philip Meier's avatar
Philip Meier committed
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
                angle=angle,
                translate=translate,
                scale=scale,
                shear=shear,
                center=center,
                interpolation=interpolation,
                fill=fill,
            )
        )

        mae = (actual.float() - expected.float()).abs().mean()
        assert mae < 2 if interpolation is transforms.InterpolationMode.NEAREST else 8

    @pytest.mark.parametrize("center", _CORRECTNESS_AFFINE_KWARGS["center"])
    @pytest.mark.parametrize(
        "interpolation", [transforms.InterpolationMode.NEAREST, transforms.InterpolationMode.BILINEAR]
    )
Philip Meier's avatar
Philip Meier committed
1092
    @pytest.mark.parametrize("fill", CORRECTNESS_FILLS)
Philip Meier's avatar
Philip Meier committed
1093
1094
    @pytest.mark.parametrize("seed", list(range(5)))
    def test_transform_image_correctness(self, center, interpolation, fill, seed):
1095
        image = make_image(dtype=torch.uint8, device="cpu")
Philip Meier's avatar
Philip Meier committed
1096

Philip Meier's avatar
Philip Meier committed
1097
        fill = adapt_fill(fill, dtype=torch.uint8)
Philip Meier's avatar
Philip Meier committed
1098
1099
1100
1101
1102
1103
1104
1105
1106

        transform = transforms.RandomAffine(
            **self._CORRECTNESS_TRANSFORM_AFFINE_RANGES, center=center, interpolation=interpolation, fill=fill
        )

        torch.manual_seed(seed)
        actual = transform(image)

        torch.manual_seed(seed)
1107
        expected = F.to_image(transform(F.to_pil_image(image)))
Philip Meier's avatar
Philip Meier committed
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133

        mae = (actual.float() - expected.float()).abs().mean()
        assert mae < 2 if interpolation is transforms.InterpolationMode.NEAREST else 8

    def _compute_affine_matrix(self, *, angle, translate, scale, shear, center):
        rot = math.radians(angle)
        cx, cy = center
        tx, ty = translate
        sx, sy = [math.radians(s) for s in ([shear, 0.0] if isinstance(shear, (int, float)) else shear)]

        c_matrix = np.array([[1, 0, cx], [0, 1, cy], [0, 0, 1]])
        t_matrix = np.array([[1, 0, tx], [0, 1, ty], [0, 0, 1]])
        c_matrix_inv = np.linalg.inv(c_matrix)
        rs_matrix = np.array(
            [
                [scale * math.cos(rot), -scale * math.sin(rot), 0],
                [scale * math.sin(rot), scale * math.cos(rot), 0],
                [0, 0, 1],
            ]
        )
        shear_x_matrix = np.array([[1, -math.tan(sx), 0], [0, 1, 0], [0, 0, 1]])
        shear_y_matrix = np.array([[1, 0, 0], [-math.tan(sy), 1, 0], [0, 0, 1]])
        rss_matrix = np.matmul(rs_matrix, np.matmul(shear_y_matrix, shear_x_matrix))
        true_matrix = np.matmul(t_matrix, np.matmul(c_matrix, np.matmul(rss_matrix, c_matrix_inv)))
        return true_matrix

1134
    def _reference_affine_bounding_boxes(self, bounding_boxes, *, angle, translate, scale, shear, center):
Philip Meier's avatar
Philip Meier committed
1135
        if center is None:
Philip Meier's avatar
Philip Meier committed
1136
            center = [s * 0.5 for s in bounding_boxes.canvas_size[::-1]]
Philip Meier's avatar
Philip Meier committed
1137
1138
1139
1140
1141
1142

        affine_matrix = self._compute_affine_matrix(
            angle=angle, translate=translate, scale=scale, shear=shear, center=center
        )
        affine_matrix = affine_matrix[:2, :]

1143
1144
1145
        expected_bboxes = reference_affine_bounding_boxes_helper(
            bounding_boxes,
            format=bounding_boxes.format,
Philip Meier's avatar
Philip Meier committed
1146
            canvas_size=bounding_boxes.canvas_size,
Philip Meier's avatar
Philip Meier committed
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
            affine_matrix=affine_matrix,
        )

        return expected_bboxes

    @pytest.mark.parametrize("format", list(datapoints.BoundingBoxFormat))
    @pytest.mark.parametrize("angle", _CORRECTNESS_AFFINE_KWARGS["angle"])
    @pytest.mark.parametrize("translate", _CORRECTNESS_AFFINE_KWARGS["translate"])
    @pytest.mark.parametrize("scale", _CORRECTNESS_AFFINE_KWARGS["scale"])
    @pytest.mark.parametrize("shear", _CORRECTNESS_AFFINE_KWARGS["shear"])
    @pytest.mark.parametrize("center", _CORRECTNESS_AFFINE_KWARGS["center"])
1158
    def test_functional_bounding_boxes_correctness(self, format, angle, translate, scale, shear, center):
1159
        bounding_boxes = make_bounding_boxes(format=format)
Philip Meier's avatar
Philip Meier committed
1160
1161

        actual = F.affine(
1162
            bounding_boxes,
Philip Meier's avatar
Philip Meier committed
1163
1164
1165
1166
1167
1168
            angle=angle,
            translate=translate,
            scale=scale,
            shear=shear,
            center=center,
        )
1169
1170
        expected = self._reference_affine_bounding_boxes(
            bounding_boxes,
Philip Meier's avatar
Philip Meier committed
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
            angle=angle,
            translate=translate,
            scale=scale,
            shear=shear,
            center=center,
        )

        torch.testing.assert_close(actual, expected)

    @pytest.mark.parametrize("format", list(datapoints.BoundingBoxFormat))
    @pytest.mark.parametrize("center", _CORRECTNESS_AFFINE_KWARGS["center"])
    @pytest.mark.parametrize("seed", list(range(5)))
1183
    def test_transform_bounding_boxes_correctness(self, format, center, seed):
1184
        bounding_boxes = make_bounding_boxes(format=format)
Philip Meier's avatar
Philip Meier committed
1185
1186
1187
1188

        transform = transforms.RandomAffine(**self._CORRECTNESS_TRANSFORM_AFFINE_RANGES, center=center)

        torch.manual_seed(seed)
1189
        params = transform._get_params([bounding_boxes])
Philip Meier's avatar
Philip Meier committed
1190
1191

        torch.manual_seed(seed)
1192
        actual = transform(bounding_boxes)
Philip Meier's avatar
Philip Meier committed
1193

1194
        expected = self._reference_affine_bounding_boxes(bounding_boxes, **params, center=center)
Philip Meier's avatar
Philip Meier committed
1195
1196
1197
1198
1199
1200
1201
1202
1203

        torch.testing.assert_close(actual, expected)

    @pytest.mark.parametrize("degrees", _EXHAUSTIVE_TYPE_TRANSFORM_AFFINE_RANGES["degrees"])
    @pytest.mark.parametrize("translate", _EXHAUSTIVE_TYPE_TRANSFORM_AFFINE_RANGES["translate"])
    @pytest.mark.parametrize("scale", _EXHAUSTIVE_TYPE_TRANSFORM_AFFINE_RANGES["scale"])
    @pytest.mark.parametrize("shear", _EXHAUSTIVE_TYPE_TRANSFORM_AFFINE_RANGES["shear"])
    @pytest.mark.parametrize("seed", list(range(10)))
    def test_transform_get_params_bounds(self, degrees, translate, scale, shear, seed):
1204
        image = make_image()
Philip Meier's avatar
Philip Meier committed
1205
        height, width = F.get_size(image)
Philip Meier's avatar
Philip Meier committed
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278

        transform = transforms.RandomAffine(degrees=degrees, translate=translate, scale=scale, shear=shear)

        torch.manual_seed(seed)
        params = transform._get_params([image])

        if isinstance(degrees, (int, float)):
            assert -degrees <= params["angle"] <= degrees
        else:
            assert degrees[0] <= params["angle"] <= degrees[1]

        if translate is not None:
            width_max = int(round(translate[0] * width))
            height_max = int(round(translate[1] * height))
            assert -width_max <= params["translate"][0] <= width_max
            assert -height_max <= params["translate"][1] <= height_max
        else:
            assert params["translate"] == (0, 0)

        if scale is not None:
            assert scale[0] <= params["scale"] <= scale[1]
        else:
            assert params["scale"] == 1.0

        if shear is not None:
            if isinstance(shear, (int, float)):
                assert -shear <= params["shear"][0] <= shear
                assert params["shear"][1] == 0.0
            elif len(shear) == 2:
                assert shear[0] <= params["shear"][0] <= shear[1]
                assert params["shear"][1] == 0.0
            elif len(shear) == 4:
                assert shear[0] <= params["shear"][0] <= shear[1]
                assert shear[2] <= params["shear"][1] <= shear[3]
        else:
            assert params["shear"] == (0, 0)

    @pytest.mark.parametrize("param", ["degrees", "translate", "scale", "shear", "center"])
    @pytest.mark.parametrize("value", [0, [0], [0, 0, 0]])
    def test_transform_sequence_len_errors(self, param, value):
        if param in {"degrees", "shear"} and not isinstance(value, list):
            return

        kwargs = {param: value}
        if param != "degrees":
            kwargs["degrees"] = 0

        with pytest.raises(
            ValueError if isinstance(value, list) else TypeError, match=f"{param} should be a sequence of length 2"
        ):
            transforms.RandomAffine(**kwargs)

    def test_transform_negative_degrees_error(self):
        with pytest.raises(ValueError, match="If degrees is a single number, it must be positive"):
            transforms.RandomAffine(degrees=-1)

    @pytest.mark.parametrize("translate", [[-1, 0], [2, 0], [-1, 2]])
    def test_transform_translate_range_error(self, translate):
        with pytest.raises(ValueError, match="translation values should be between 0 and 1"):
            transforms.RandomAffine(degrees=0, translate=translate)

    @pytest.mark.parametrize("scale", [[-1, 0], [0, -1], [-1, -1]])
    def test_transform_scale_range_error(self, scale):
        with pytest.raises(ValueError, match="scale values should be positive"):
            transforms.RandomAffine(degrees=0, scale=scale)

    def test_transform_negative_shear_error(self):
        with pytest.raises(ValueError, match="If shear is a single number, it must be positive"):
            transforms.RandomAffine(degrees=0, shear=-1)

    def test_transform_unknown_fill_error(self):
        with pytest.raises(TypeError, match="Got inappropriate fill arg"):
            transforms.RandomAffine(degrees=0, fill="fill")
Philip Meier's avatar
Philip Meier committed
1279
1280
1281
1282
1283
1284


class TestVerticalFlip:
    @pytest.mark.parametrize("dtype", [torch.float32, torch.uint8])
    @pytest.mark.parametrize("device", cpu_and_cuda())
    def test_kernel_image_tensor(self, dtype, device):
1285
        check_kernel(F.vertical_flip_image, make_image(dtype=dtype, device=device))
Philip Meier's avatar
Philip Meier committed
1286
1287
1288
1289

    @pytest.mark.parametrize("format", list(datapoints.BoundingBoxFormat))
    @pytest.mark.parametrize("dtype", [torch.float32, torch.int64])
    @pytest.mark.parametrize("device", cpu_and_cuda())
1290
    def test_kernel_bounding_boxes(self, format, dtype, device):
1291
        bounding_boxes = make_bounding_boxes(format=format, dtype=dtype, device=device)
Philip Meier's avatar
Philip Meier committed
1292
        check_kernel(
1293
1294
            F.vertical_flip_bounding_boxes,
            bounding_boxes,
Philip Meier's avatar
Philip Meier committed
1295
            format=format,
Philip Meier's avatar
Philip Meier committed
1296
            canvas_size=bounding_boxes.canvas_size,
Philip Meier's avatar
Philip Meier committed
1297
1298
        )

1299
1300
1301
    @pytest.mark.parametrize("make_mask", [make_segmentation_mask, make_detection_mask])
    def test_kernel_mask(self, make_mask):
        check_kernel(F.vertical_flip_mask, make_mask())
Philip Meier's avatar
Philip Meier committed
1302
1303

    def test_kernel_video(self):
1304
        check_kernel(F.vertical_flip_video, make_video())
Philip Meier's avatar
Philip Meier committed
1305
1306

    @pytest.mark.parametrize(
Philip Meier's avatar
Philip Meier committed
1307
        "make_input",
1308
        [make_image_tensor, make_image_pil, make_image, make_bounding_boxes, make_segmentation_mask, make_video],
Philip Meier's avatar
Philip Meier committed
1309
    )
Nicolas Hug's avatar
Nicolas Hug committed
1310
1311
    def test_functional(self, make_input):
        check_functional(F.vertical_flip, make_input())
Philip Meier's avatar
Philip Meier committed
1312
1313

    @pytest.mark.parametrize(
1314
        ("kernel", "input_type"),
Philip Meier's avatar
Philip Meier committed
1315
        [
1316
1317
1318
            (F.vertical_flip_image, torch.Tensor),
            (F._vertical_flip_image_pil, PIL.Image.Image),
            (F.vertical_flip_image, datapoints.Image),
1319
            (F.vertical_flip_bounding_boxes, datapoints.BoundingBoxes),
1320
1321
            (F.vertical_flip_mask, datapoints.Mask),
            (F.vertical_flip_video, datapoints.Video),
Philip Meier's avatar
Philip Meier committed
1322
1323
        ],
    )
Nicolas Hug's avatar
Nicolas Hug committed
1324
1325
    def test_functional_signature(self, kernel, input_type):
        check_functional_kernel_signature_match(F.vertical_flip, kernel=kernel, input_type=input_type)
Philip Meier's avatar
Philip Meier committed
1326
1327

    @pytest.mark.parametrize(
1328
        "make_input",
1329
        [make_image_tensor, make_image_pil, make_image, make_bounding_boxes, make_segmentation_mask, make_video],
Philip Meier's avatar
Philip Meier committed
1330
1331
    )
    @pytest.mark.parametrize("device", cpu_and_cuda())
1332
    def test_transform(self, make_input, device):
1333
        check_transform(transforms.RandomVerticalFlip(p=1), make_input(device=device))
Philip Meier's avatar
Philip Meier committed
1334
1335
1336

    @pytest.mark.parametrize("fn", [F.vertical_flip, transform_cls_to_functional(transforms.RandomVerticalFlip, p=1)])
    def test_image_correctness(self, fn):
1337
        image = make_image(dtype=torch.uint8, device="cpu")
Philip Meier's avatar
Philip Meier committed
1338
1339

        actual = fn(image)
1340
        expected = F.to_image(F.vertical_flip(F.to_pil_image(image)))
Philip Meier's avatar
Philip Meier committed
1341
1342
1343

        torch.testing.assert_close(actual, expected)

1344
    def _reference_vertical_flip_bounding_boxes(self, bounding_boxes):
Philip Meier's avatar
Philip Meier committed
1345
1346
1347
        affine_matrix = np.array(
            [
                [1, 0, 0],
Philip Meier's avatar
Philip Meier committed
1348
                [0, -1, bounding_boxes.canvas_size[0]],
Philip Meier's avatar
Philip Meier committed
1349
            ],
1350
            dtype="float64" if bounding_boxes.dtype == torch.float64 else "float32",
Philip Meier's avatar
Philip Meier committed
1351
1352
        )

1353
1354
1355
        expected_bboxes = reference_affine_bounding_boxes_helper(
            bounding_boxes,
            format=bounding_boxes.format,
Philip Meier's avatar
Philip Meier committed
1356
            canvas_size=bounding_boxes.canvas_size,
Philip Meier's avatar
Philip Meier committed
1357
1358
1359
            affine_matrix=affine_matrix,
        )

1360
        return datapoints.wrap(expected_bboxes, like=bounding_boxes)
Philip Meier's avatar
Philip Meier committed
1361
1362
1363

    @pytest.mark.parametrize("format", list(datapoints.BoundingBoxFormat))
    @pytest.mark.parametrize("fn", [F.vertical_flip, transform_cls_to_functional(transforms.RandomVerticalFlip, p=1)])
1364
    def test_bounding_boxes_correctness(self, format, fn):
1365
        bounding_boxes = make_bounding_boxes(format=format)
Philip Meier's avatar
Philip Meier committed
1366

1367
1368
        actual = fn(bounding_boxes)
        expected = self._reference_vertical_flip_bounding_boxes(bounding_boxes)
Philip Meier's avatar
Philip Meier committed
1369
1370
1371
1372

        torch.testing.assert_close(actual, expected)

    @pytest.mark.parametrize(
1373
        "make_input",
1374
        [make_image_tensor, make_image_pil, make_image, make_bounding_boxes, make_segmentation_mask, make_video],
Philip Meier's avatar
Philip Meier committed
1375
1376
    )
    @pytest.mark.parametrize("device", cpu_and_cuda())
1377
1378
    def test_transform_noop(self, make_input, device):
        input = make_input(device=device)
Philip Meier's avatar
Philip Meier committed
1379
1380
1381
1382
1383
1384

        transform = transforms.RandomVerticalFlip(p=0)

        output = transform(input)

        assert_equal(output, input)
Philip Meier's avatar
Philip Meier committed
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419


class TestRotate:
    _EXHAUSTIVE_TYPE_AFFINE_KWARGS = dict(
        # float, int
        angle=[-10.9, 18],
        # None
        # two-list of float, two-list of int, two-tuple of float, two-tuple of int
        center=[None, [1.2, 4.9], [-3, 1], (2.5, -4.7), (3, 2)],
    )
    _MINIMAL_AFFINE_KWARGS = {k: vs[0] for k, vs in _EXHAUSTIVE_TYPE_AFFINE_KWARGS.items()}
    _CORRECTNESS_AFFINE_KWARGS = {
        k: [v for v in vs if v is None or isinstance(v, float) or isinstance(v, list)]
        for k, vs in _EXHAUSTIVE_TYPE_AFFINE_KWARGS.items()
    }

    _EXHAUSTIVE_TYPE_TRANSFORM_AFFINE_RANGES = dict(
        degrees=[30, (-15, 20)],
    )
    _CORRECTNESS_TRANSFORM_AFFINE_RANGES = {k: vs[0] for k, vs in _EXHAUSTIVE_TYPE_TRANSFORM_AFFINE_RANGES.items()}

    @param_value_parametrization(
        angle=_EXHAUSTIVE_TYPE_AFFINE_KWARGS["angle"],
        interpolation=[transforms.InterpolationMode.NEAREST, transforms.InterpolationMode.BILINEAR],
        expand=[False, True],
        center=_EXHAUSTIVE_TYPE_AFFINE_KWARGS["center"],
        fill=EXHAUSTIVE_TYPE_FILLS,
    )
    @pytest.mark.parametrize("dtype", [torch.float32, torch.uint8])
    @pytest.mark.parametrize("device", cpu_and_cuda())
    def test_kernel_image_tensor(self, param, value, dtype, device):
        kwargs = {param: value}
        if param != "angle":
            kwargs["angle"] = self._MINIMAL_AFFINE_KWARGS["angle"]
        check_kernel(
1420
            F.rotate_image,
1421
            make_image(dtype=dtype, device=device),
Philip Meier's avatar
Philip Meier committed
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
            **kwargs,
            check_scripted_vs_eager=not (param == "fill" and isinstance(value, (int, float))),
        )

    @param_value_parametrization(
        angle=_EXHAUSTIVE_TYPE_AFFINE_KWARGS["angle"],
        expand=[False, True],
        center=_EXHAUSTIVE_TYPE_AFFINE_KWARGS["center"],
    )
    @pytest.mark.parametrize("format", list(datapoints.BoundingBoxFormat))
    @pytest.mark.parametrize("dtype", [torch.float32, torch.uint8])
    @pytest.mark.parametrize("device", cpu_and_cuda())
1434
    def test_kernel_bounding_boxes(self, param, value, format, dtype, device):
Philip Meier's avatar
Philip Meier committed
1435
1436
1437
1438
        kwargs = {param: value}
        if param != "angle":
            kwargs["angle"] = self._MINIMAL_AFFINE_KWARGS["angle"]

1439
        bounding_boxes = make_bounding_boxes(format=format, dtype=dtype, device=device)
Philip Meier's avatar
Philip Meier committed
1440
1441

        check_kernel(
1442
1443
            F.rotate_bounding_boxes,
            bounding_boxes,
Philip Meier's avatar
Philip Meier committed
1444
            format=format,
Philip Meier's avatar
Philip Meier committed
1445
            canvas_size=bounding_boxes.canvas_size,
Philip Meier's avatar
Philip Meier committed
1446
1447
1448
            **kwargs,
        )

1449
1450
1451
    @pytest.mark.parametrize("make_mask", [make_segmentation_mask, make_detection_mask])
    def test_kernel_mask(self, make_mask):
        check_kernel(F.rotate_mask, make_mask(), **self._MINIMAL_AFFINE_KWARGS)
Philip Meier's avatar
Philip Meier committed
1452
1453

    def test_kernel_video(self):
1454
        check_kernel(F.rotate_video, make_video(), **self._MINIMAL_AFFINE_KWARGS)
Philip Meier's avatar
Philip Meier committed
1455
1456

    @pytest.mark.parametrize(
Philip Meier's avatar
Philip Meier committed
1457
        "make_input",
1458
        [make_image_tensor, make_image_pil, make_image, make_bounding_boxes, make_segmentation_mask, make_video],
Philip Meier's avatar
Philip Meier committed
1459
    )
Nicolas Hug's avatar
Nicolas Hug committed
1460
1461
    def test_functional(self, make_input):
        check_functional(F.rotate, make_input(), **self._MINIMAL_AFFINE_KWARGS)
Philip Meier's avatar
Philip Meier committed
1462
1463

    @pytest.mark.parametrize(
1464
        ("kernel", "input_type"),
Philip Meier's avatar
Philip Meier committed
1465
        [
1466
1467
1468
            (F.rotate_image, torch.Tensor),
            (F._rotate_image_pil, PIL.Image.Image),
            (F.rotate_image, datapoints.Image),
1469
            (F.rotate_bounding_boxes, datapoints.BoundingBoxes),
1470
1471
            (F.rotate_mask, datapoints.Mask),
            (F.rotate_video, datapoints.Video),
Philip Meier's avatar
Philip Meier committed
1472
1473
        ],
    )
Nicolas Hug's avatar
Nicolas Hug committed
1474
1475
    def test_functional_signature(self, kernel, input_type):
        check_functional_kernel_signature_match(F.rotate, kernel=kernel, input_type=input_type)
Philip Meier's avatar
Philip Meier committed
1476
1477

    @pytest.mark.parametrize(
1478
        "make_input",
1479
        [make_image_tensor, make_image_pil, make_image, make_bounding_boxes, make_segmentation_mask, make_video],
Philip Meier's avatar
Philip Meier committed
1480
1481
    )
    @pytest.mark.parametrize("device", cpu_and_cuda())
1482
1483
    def test_transform(self, make_input, device):
        check_transform(
1484
            transforms.RandomRotation(**self._CORRECTNESS_TRANSFORM_AFFINE_RANGES), make_input(device=device)
1485
        )
Philip Meier's avatar
Philip Meier committed
1486
1487
1488
1489
1490
1491
1492
1493
1494

    @pytest.mark.parametrize("angle", _CORRECTNESS_AFFINE_KWARGS["angle"])
    @pytest.mark.parametrize("center", _CORRECTNESS_AFFINE_KWARGS["center"])
    @pytest.mark.parametrize(
        "interpolation", [transforms.InterpolationMode.NEAREST, transforms.InterpolationMode.BILINEAR]
    )
    @pytest.mark.parametrize("expand", [False, True])
    @pytest.mark.parametrize("fill", CORRECTNESS_FILLS)
    def test_functional_image_correctness(self, angle, center, interpolation, expand, fill):
1495
        image = make_image(dtype=torch.uint8, device="cpu")
Philip Meier's avatar
Philip Meier committed
1496
1497
1498
1499

        fill = adapt_fill(fill, dtype=torch.uint8)

        actual = F.rotate(image, angle=angle, center=center, interpolation=interpolation, expand=expand, fill=fill)
1500
        expected = F.to_image(
Philip Meier's avatar
Philip Meier committed
1501
            F.rotate(
1502
                F.to_pil_image(image), angle=angle, center=center, interpolation=interpolation, expand=expand, fill=fill
Philip Meier's avatar
Philip Meier committed
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
            )
        )

        mae = (actual.float() - expected.float()).abs().mean()
        assert mae < 1 if interpolation is transforms.InterpolationMode.NEAREST else 6

    @pytest.mark.parametrize("center", _CORRECTNESS_AFFINE_KWARGS["center"])
    @pytest.mark.parametrize(
        "interpolation", [transforms.InterpolationMode.NEAREST, transforms.InterpolationMode.BILINEAR]
    )
    @pytest.mark.parametrize("expand", [False, True])
    @pytest.mark.parametrize("fill", CORRECTNESS_FILLS)
    @pytest.mark.parametrize("seed", list(range(5)))
    def test_transform_image_correctness(self, center, interpolation, expand, fill, seed):
1517
        image = make_image(dtype=torch.uint8, device="cpu")
Philip Meier's avatar
Philip Meier committed
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532

        fill = adapt_fill(fill, dtype=torch.uint8)

        transform = transforms.RandomRotation(
            **self._CORRECTNESS_TRANSFORM_AFFINE_RANGES,
            center=center,
            interpolation=interpolation,
            expand=expand,
            fill=fill,
        )

        torch.manual_seed(seed)
        actual = transform(image)

        torch.manual_seed(seed)
1533
        expected = F.to_image(transform(F.to_pil_image(image)))
Philip Meier's avatar
Philip Meier committed
1534
1535
1536
1537

        mae = (actual.float() - expected.float()).abs().mean()
        assert mae < 1 if interpolation is transforms.InterpolationMode.NEAREST else 6

1538
    def _reference_rotate_bounding_boxes(self, bounding_boxes, *, angle, expand, center):
Philip Meier's avatar
Philip Meier committed
1539
1540
1541
1542
1543
        # FIXME
        if expand:
            raise ValueError("This reference currently does not support expand=True")

        if center is None:
Philip Meier's avatar
Philip Meier committed
1544
            center = [s * 0.5 for s in bounding_boxes.canvas_size[::-1]]
Philip Meier's avatar
Philip Meier committed
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554

        a = np.cos(angle * np.pi / 180.0)
        b = np.sin(angle * np.pi / 180.0)
        cx = center[0]
        cy = center[1]
        affine_matrix = np.array(
            [
                [a, b, cx - cx * a - b * cy],
                [-b, a, cy + cx * b - a * cy],
            ],
1555
            dtype="float64" if bounding_boxes.dtype == torch.float64 else "float32",
Philip Meier's avatar
Philip Meier committed
1556
1557
        )

1558
1559
1560
        expected_bboxes = reference_affine_bounding_boxes_helper(
            bounding_boxes,
            format=bounding_boxes.format,
Philip Meier's avatar
Philip Meier committed
1561
            canvas_size=bounding_boxes.canvas_size,
Philip Meier's avatar
Philip Meier committed
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
            affine_matrix=affine_matrix,
        )

        return expected_bboxes

    @pytest.mark.parametrize("format", list(datapoints.BoundingBoxFormat))
    @pytest.mark.parametrize("angle", _CORRECTNESS_AFFINE_KWARGS["angle"])
    # TODO: add support for expand=True in the reference
    @pytest.mark.parametrize("expand", [False])
    @pytest.mark.parametrize("center", _CORRECTNESS_AFFINE_KWARGS["center"])
1572
    def test_functional_bounding_boxes_correctness(self, format, angle, expand, center):
1573
        bounding_boxes = make_bounding_boxes(format=format)
Philip Meier's avatar
Philip Meier committed
1574

1575
1576
        actual = F.rotate(bounding_boxes, angle=angle, expand=expand, center=center)
        expected = self._reference_rotate_bounding_boxes(bounding_boxes, angle=angle, expand=expand, center=center)
Philip Meier's avatar
Philip Meier committed
1577
1578
1579
1580
1581
1582
1583
1584

        torch.testing.assert_close(actual, expected)

    @pytest.mark.parametrize("format", list(datapoints.BoundingBoxFormat))
    # TODO: add support for expand=True in the reference
    @pytest.mark.parametrize("expand", [False])
    @pytest.mark.parametrize("center", _CORRECTNESS_AFFINE_KWARGS["center"])
    @pytest.mark.parametrize("seed", list(range(5)))
1585
    def test_transform_bounding_boxes_correctness(self, format, expand, center, seed):
1586
        bounding_boxes = make_bounding_boxes(format=format)
Philip Meier's avatar
Philip Meier committed
1587
1588
1589
1590

        transform = transforms.RandomRotation(**self._CORRECTNESS_TRANSFORM_AFFINE_RANGES, expand=expand, center=center)

        torch.manual_seed(seed)
1591
        params = transform._get_params([bounding_boxes])
Philip Meier's avatar
Philip Meier committed
1592
1593

        torch.manual_seed(seed)
1594
        actual = transform(bounding_boxes)
Philip Meier's avatar
Philip Meier committed
1595

1596
        expected = self._reference_rotate_bounding_boxes(bounding_boxes, **params, expand=expand, center=center)
Philip Meier's avatar
Philip Meier committed
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634

        torch.testing.assert_close(actual, expected)

    @pytest.mark.parametrize("degrees", _EXHAUSTIVE_TYPE_TRANSFORM_AFFINE_RANGES["degrees"])
    @pytest.mark.parametrize("seed", list(range(10)))
    def test_transform_get_params_bounds(self, degrees, seed):
        transform = transforms.RandomRotation(degrees=degrees)

        torch.manual_seed(seed)
        params = transform._get_params([])

        if isinstance(degrees, (int, float)):
            assert -degrees <= params["angle"] <= degrees
        else:
            assert degrees[0] <= params["angle"] <= degrees[1]

    @pytest.mark.parametrize("param", ["degrees", "center"])
    @pytest.mark.parametrize("value", [0, [0], [0, 0, 0]])
    def test_transform_sequence_len_errors(self, param, value):
        if param == "degrees" and not isinstance(value, list):
            return

        kwargs = {param: value}
        if param != "degrees":
            kwargs["degrees"] = 0

        with pytest.raises(
            ValueError if isinstance(value, list) else TypeError, match=f"{param} should be a sequence of length 2"
        ):
            transforms.RandomRotation(**kwargs)

    def test_transform_negative_degrees_error(self):
        with pytest.raises(ValueError, match="If degrees is a single number, it must be positive"):
            transforms.RandomAffine(degrees=-1)

    def test_transform_unknown_fill_error(self):
        with pytest.raises(TypeError, match="Got inappropriate fill arg"):
            transforms.RandomAffine(degrees=0, fill="fill")
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
1655
1656
1657
1658
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
1695


class TestCompose:
    class BuiltinTransform(transforms.Transform):
        def _transform(self, inpt, params):
            return inpt

    class PackedInputTransform(nn.Module):
        def forward(self, sample):
            assert len(sample) == 2
            return sample

    class UnpackedInputTransform(nn.Module):
        def forward(self, image, label):
            return image, label

    @pytest.mark.parametrize(
        "transform_clss",
        [
            [BuiltinTransform],
            [PackedInputTransform],
            [UnpackedInputTransform],
            [BuiltinTransform, BuiltinTransform],
            [PackedInputTransform, PackedInputTransform],
            [UnpackedInputTransform, UnpackedInputTransform],
            [BuiltinTransform, PackedInputTransform, BuiltinTransform],
            [BuiltinTransform, UnpackedInputTransform, BuiltinTransform],
            [PackedInputTransform, BuiltinTransform, PackedInputTransform],
            [UnpackedInputTransform, BuiltinTransform, UnpackedInputTransform],
        ],
    )
    @pytest.mark.parametrize("unpack", [True, False])
    def test_packed_unpacked(self, transform_clss, unpack):
        needs_packed_inputs = any(issubclass(cls, self.PackedInputTransform) for cls in transform_clss)
        needs_unpacked_inputs = any(issubclass(cls, self.UnpackedInputTransform) for cls in transform_clss)
        assert not (needs_packed_inputs and needs_unpacked_inputs)

        transform = transforms.Compose([cls() for cls in transform_clss])

        image = make_image()
        label = 3
        packed_input = (image, label)

        def call_transform():
            if unpack:
                return transform(*packed_input)
            else:
                return transform(packed_input)

        if needs_unpacked_inputs and not unpack:
            with pytest.raises(TypeError, match="missing 1 required positional argument"):
                call_transform()
        elif needs_packed_inputs and unpack:
            with pytest.raises(TypeError, match="takes 2 positional arguments but 3 were given"):
                call_transform()
        else:
            output = call_transform()

            assert isinstance(output, tuple) and len(output) == 2
            assert output[0] is image
            assert output[1] is label
1696
1697
1698
1699
1700
1701


class TestToDtype:
    @pytest.mark.parametrize(
        ("kernel", "make_input"),
        [
1702
1703
            (F.to_dtype_image, make_image_tensor),
            (F.to_dtype_image, make_image),
1704
1705
1706
1707
1708
1709
1710
1711
1712
1713
1714
1715
1716
1717
1718
1719
            (F.to_dtype_video, make_video),
        ],
    )
    @pytest.mark.parametrize("input_dtype", [torch.float32, torch.float64, torch.uint8])
    @pytest.mark.parametrize("output_dtype", [torch.float32, torch.float64, torch.uint8])
    @pytest.mark.parametrize("device", cpu_and_cuda())
    @pytest.mark.parametrize("scale", (True, False))
    def test_kernel(self, kernel, make_input, input_dtype, output_dtype, device, scale):
        check_kernel(
            kernel,
            make_input(dtype=input_dtype, device=device),
            expect_same_dtype=input_dtype is output_dtype,
            dtype=output_dtype,
            scale=scale,
        )

Philip Meier's avatar
Philip Meier committed
1720
    @pytest.mark.parametrize("make_input", [make_image_tensor, make_image, make_video])
1721
1722
1723
1724
    @pytest.mark.parametrize("input_dtype", [torch.float32, torch.float64, torch.uint8])
    @pytest.mark.parametrize("output_dtype", [torch.float32, torch.float64, torch.uint8])
    @pytest.mark.parametrize("device", cpu_and_cuda())
    @pytest.mark.parametrize("scale", (True, False))
Nicolas Hug's avatar
Nicolas Hug committed
1725
1726
    def test_functional(self, make_input, input_dtype, output_dtype, device, scale):
        check_functional(
1727
1728
1729
1730
1731
1732
1733
1734
            F.to_dtype,
            make_input(dtype=input_dtype, device=device),
            dtype=output_dtype,
            scale=scale,
        )

    @pytest.mark.parametrize(
        "make_input",
1735
        [make_image_tensor, make_image, make_bounding_boxes, make_segmentation_mask, make_video],
1736
1737
1738
1739
1740
1741
1742
1743
1744
1745
    )
    @pytest.mark.parametrize("input_dtype", [torch.float32, torch.float64, torch.uint8])
    @pytest.mark.parametrize("output_dtype", [torch.float32, torch.float64, torch.uint8])
    @pytest.mark.parametrize("device", cpu_and_cuda())
    @pytest.mark.parametrize("scale", (True, False))
    @pytest.mark.parametrize("as_dict", (True, False))
    def test_transform(self, make_input, input_dtype, output_dtype, device, scale, as_dict):
        input = make_input(dtype=input_dtype, device=device)
        if as_dict:
            output_dtype = {type(input): output_dtype}
1746
        check_transform(transforms.ToDtype(dtype=output_dtype, scale=scale), input)
1747
1748
1749
1750
1751
1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766
1767
1768
1769
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791
1792
1793
1794
1795
1796
1797
1798
1799
1800
1801
1802
1803
1804
1805
1806
1807
1808
1809

    def reference_convert_dtype_image_tensor(self, image, dtype=torch.float, scale=False):
        input_dtype = image.dtype
        output_dtype = dtype

        if not scale:
            return image.to(dtype)

        if output_dtype == input_dtype:
            return image

        def fn(value):
            if input_dtype.is_floating_point:
                if output_dtype.is_floating_point:
                    return value
                else:
                    return round(decimal.Decimal(value) * torch.iinfo(output_dtype).max)
            else:
                input_max_value = torch.iinfo(input_dtype).max

                if output_dtype.is_floating_point:
                    return float(decimal.Decimal(value) / input_max_value)
                else:
                    output_max_value = torch.iinfo(output_dtype).max

                    if input_max_value > output_max_value:
                        factor = (input_max_value + 1) // (output_max_value + 1)
                        return value / factor
                    else:
                        factor = (output_max_value + 1) // (input_max_value + 1)
                        return value * factor

        return torch.tensor(tree_map(fn, image.tolist()), dtype=dtype, device=image.device)

    @pytest.mark.parametrize("input_dtype", [torch.float32, torch.float64, torch.uint8])
    @pytest.mark.parametrize("output_dtype", [torch.float32, torch.float64, torch.uint8])
    @pytest.mark.parametrize("device", cpu_and_cuda())
    @pytest.mark.parametrize("scale", (True, False))
    def test_image_correctness(self, input_dtype, output_dtype, device, scale):
        if input_dtype.is_floating_point and output_dtype == torch.int64:
            pytest.xfail("float to int64 conversion is not supported")

        input = make_image(dtype=input_dtype, device=device)

        out = F.to_dtype(input, dtype=output_dtype, scale=scale)
        expected = self.reference_convert_dtype_image_tensor(input, dtype=output_dtype, scale=scale)

        if input_dtype.is_floating_point and not output_dtype.is_floating_point and scale:
            torch.testing.assert_close(out, expected, atol=1, rtol=0)
        else:
            torch.testing.assert_close(out, expected)

    def was_scaled(self, inpt):
        # this assumes the target dtype is float
        return inpt.max() <= 1

    def make_inpt_with_bbox_and_mask(self, make_input):
        H, W = 10, 10
        inpt_dtype = torch.uint8
        bbox_dtype = torch.float32
        mask_dtype = torch.bool
        sample = {
            "inpt": make_input(size=(H, W), dtype=inpt_dtype),
1810
            "bbox": make_bounding_boxes(canvas_size=(H, W), dtype=bbox_dtype),
1811
1812
1813
1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
1824
1825
1826
1827
1828
1829
1830
1831
1832
1833
1834
1835
1836
1837
1838
1839
1840
1841
1842
1843
1844
1845
1846
1847
1848
1849
1850
1851
1852
1853
1854
1855
1856
1857
1858
1859
1860
1861
1862
1863
1864
1865
1866
1867
1868
1869
1870
1871
1872
            "mask": make_detection_mask(size=(H, W), dtype=mask_dtype),
        }

        return sample, inpt_dtype, bbox_dtype, mask_dtype

    @pytest.mark.parametrize("make_input", (make_image_tensor, make_image, make_video))
    @pytest.mark.parametrize("scale", (True, False))
    def test_dtype_not_a_dict(self, make_input, scale):
        # assert only inpt gets transformed when dtype isn't a dict

        sample, inpt_dtype, bbox_dtype, mask_dtype = self.make_inpt_with_bbox_and_mask(make_input)
        out = transforms.ToDtype(dtype=torch.float32, scale=scale)(sample)

        assert out["inpt"].dtype != inpt_dtype
        assert out["inpt"].dtype == torch.float32
        if scale:
            assert self.was_scaled(out["inpt"])
        else:
            assert not self.was_scaled(out["inpt"])
        assert out["bbox"].dtype == bbox_dtype
        assert out["mask"].dtype == mask_dtype

    @pytest.mark.parametrize("make_input", (make_image_tensor, make_image, make_video))
    def test_others_catch_all_and_none(self, make_input):
        # make sure "others" works as a catch-all and that None means no conversion

        sample, inpt_dtype, bbox_dtype, mask_dtype = self.make_inpt_with_bbox_and_mask(make_input)
        out = transforms.ToDtype(dtype={datapoints.Mask: torch.int64, "others": None})(sample)
        assert out["inpt"].dtype == inpt_dtype
        assert out["bbox"].dtype == bbox_dtype
        assert out["mask"].dtype != mask_dtype
        assert out["mask"].dtype == torch.int64

    @pytest.mark.parametrize("make_input", (make_image_tensor, make_image, make_video))
    def test_typical_use_case(self, make_input):
        # Typical use-case: want to convert dtype and scale for inpt and just dtype for masks.
        # This just makes sure we now have a decent API for this

        sample, inpt_dtype, bbox_dtype, mask_dtype = self.make_inpt_with_bbox_and_mask(make_input)
        out = transforms.ToDtype(
            dtype={type(sample["inpt"]): torch.float32, datapoints.Mask: torch.int64, "others": None}, scale=True
        )(sample)
        assert out["inpt"].dtype != inpt_dtype
        assert out["inpt"].dtype == torch.float32
        assert self.was_scaled(out["inpt"])
        assert out["bbox"].dtype == bbox_dtype
        assert out["mask"].dtype != mask_dtype
        assert out["mask"].dtype == torch.int64

    @pytest.mark.parametrize("make_input", (make_image_tensor, make_image, make_video))
    def test_errors_warnings(self, make_input):
        sample, inpt_dtype, bbox_dtype, mask_dtype = self.make_inpt_with_bbox_and_mask(make_input)

        with pytest.raises(ValueError, match="No dtype was specified for"):
            out = transforms.ToDtype(dtype={datapoints.Mask: torch.float32})(sample)
        with pytest.warns(UserWarning, match=re.escape("plain `torch.Tensor` will *not* be transformed")):
            transforms.ToDtype(dtype={torch.Tensor: torch.float32, datapoints.Image: torch.float32})
        with pytest.warns(UserWarning, match="no scaling will be done"):
            out = transforms.ToDtype(dtype={"others": None}, scale=True)(sample)
        assert out["inpt"].dtype == inpt_dtype
        assert out["bbox"].dtype == bbox_dtype
        assert out["mask"].dtype == mask_dtype
1873
1874


1875
1876
1877
1878
1879
1880
1881
class TestAdjustBrightness:
    _CORRECTNESS_BRIGHTNESS_FACTORS = [0.5, 0.0, 1.0, 5.0]
    _DEFAULT_BRIGHTNESS_FACTOR = _CORRECTNESS_BRIGHTNESS_FACTORS[0]

    @pytest.mark.parametrize(
        ("kernel", "make_input"),
        [
1882
            (F.adjust_brightness_image, make_image),
1883
1884
1885
1886
1887
1888
1889
1890
            (F.adjust_brightness_video, make_video),
        ],
    )
    @pytest.mark.parametrize("dtype", [torch.float32, torch.uint8])
    @pytest.mark.parametrize("device", cpu_and_cuda())
    def test_kernel(self, kernel, make_input, dtype, device):
        check_kernel(kernel, make_input(dtype=dtype, device=device), brightness_factor=self._DEFAULT_BRIGHTNESS_FACTOR)

Philip Meier's avatar
Philip Meier committed
1891
    @pytest.mark.parametrize("make_input", [make_image_tensor, make_image_pil, make_image, make_video])
Nicolas Hug's avatar
Nicolas Hug committed
1892
1893
    def test_functional(self, make_input):
        check_functional(F.adjust_brightness, make_input(), brightness_factor=self._DEFAULT_BRIGHTNESS_FACTOR)
1894
1895
1896
1897

    @pytest.mark.parametrize(
        ("kernel", "input_type"),
        [
1898
1899
1900
            (F.adjust_brightness_image, torch.Tensor),
            (F._adjust_brightness_image_pil, PIL.Image.Image),
            (F.adjust_brightness_image, datapoints.Image),
1901
1902
1903
            (F.adjust_brightness_video, datapoints.Video),
        ],
    )
Nicolas Hug's avatar
Nicolas Hug committed
1904
1905
    def test_functional_signature(self, kernel, input_type):
        check_functional_kernel_signature_match(F.adjust_brightness, kernel=kernel, input_type=input_type)
1906
1907
1908
1909
1910
1911

    @pytest.mark.parametrize("brightness_factor", _CORRECTNESS_BRIGHTNESS_FACTORS)
    def test_image_correctness(self, brightness_factor):
        image = make_image(dtype=torch.uint8, device="cpu")

        actual = F.adjust_brightness(image, brightness_factor=brightness_factor)
1912
        expected = F.to_image(F.adjust_brightness(F.to_pil_image(image), brightness_factor=brightness_factor))
1913
1914
1915
1916

        torch.testing.assert_close(actual, expected)


1917
1918
1919
1920
1921
1922
1923
1924
1925
1926
1927
1928
1929
1930
1931
class TestCutMixMixUp:
    class DummyDataset:
        def __init__(self, size, num_classes):
            self.size = size
            self.num_classes = num_classes
            assert size < num_classes

        def __getitem__(self, idx):
            img = torch.rand(3, 100, 100)
            label = idx  # This ensures all labels in a batch are unique and makes testing easier
            return img, label

        def __len__(self):
            return self.size

Nicolas Hug's avatar
Nicolas Hug committed
1932
    @pytest.mark.parametrize("T", [transforms.CutMix, transforms.MixUp])
1933
1934
1935
1936
1937
1938
1939
    def test_supported_input_structure(self, T):

        batch_size = 32
        num_classes = 100

        dataset = self.DummyDataset(size=batch_size, num_classes=num_classes)

1940
        cutmix_mixup = T(num_classes=num_classes)
1941
1942
1943
1944
1945
1946
1947
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981

        dl = DataLoader(dataset, batch_size=batch_size)

        # Input sanity checks
        img, target = next(iter(dl))
        input_img_size = img.shape[-3:]
        assert isinstance(img, torch.Tensor) and isinstance(target, torch.Tensor)
        assert target.shape == (batch_size,)

        def check_output(img, target):
            assert img.shape == (batch_size, *input_img_size)
            assert target.shape == (batch_size, num_classes)
            torch.testing.assert_close(target.sum(axis=-1), torch.ones(batch_size))
            num_non_zero_labels = (target != 0).sum(axis=-1)
            assert (num_non_zero_labels == 2).all()

        # After Dataloader, as unpacked input
        img, target = next(iter(dl))
        assert target.shape == (batch_size,)
        img, target = cutmix_mixup(img, target)
        check_output(img, target)

        # After Dataloader, as packed input
        packed_from_dl = next(iter(dl))
        assert isinstance(packed_from_dl, list)
        img, target = cutmix_mixup(packed_from_dl)
        check_output(img, target)

        # As collation function. We expect default_collate to be used by users.
        def collate_fn_1(batch):
            return cutmix_mixup(default_collate(batch))

        def collate_fn_2(batch):
            return cutmix_mixup(*default_collate(batch))

        for collate_fn in (collate_fn_1, collate_fn_2):
            dl = DataLoader(dataset, batch_size=batch_size, collate_fn=collate_fn)
            img, target = next(iter(dl))
            check_output(img, target)

    @needs_cuda
Nicolas Hug's avatar
Nicolas Hug committed
1982
    @pytest.mark.parametrize("T", [transforms.CutMix, transforms.MixUp])
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
    def test_cpu_vs_gpu(self, T):
        num_classes = 10
        batch_size = 3
        H, W = 12, 12

        imgs = torch.rand(batch_size, 3, H, W)
        labels = torch.randint(0, num_classes, (batch_size,))
        cutmix_mixup = T(alpha=0.5, num_classes=num_classes)

        _check_kernel_cuda_vs_cpu(cutmix_mixup, imgs, labels, rtol=None, atol=None)

Nicolas Hug's avatar
Nicolas Hug committed
1994
    @pytest.mark.parametrize("T", [transforms.CutMix, transforms.MixUp])
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
    def test_error(self, T):

        num_classes = 10
        batch_size = 9

        imgs = torch.rand(batch_size, 3, 12, 12)
        cutmix_mixup = T(alpha=0.5, num_classes=num_classes)

        for input_with_bad_type in (
            F.to_pil_image(imgs[0]),
            datapoints.Mask(torch.rand(12, 12)),
Philip Meier's avatar
Philip Meier committed
2006
            datapoints.BoundingBoxes(torch.rand(2, 4), format="XYXY", canvas_size=12),
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
2036
2037
2038
2039
2040
2041
2042
2043
2044
2045
2046
2047
2048
2049
2050
2051
2052
2053
        ):
            with pytest.raises(ValueError, match="does not support PIL images, "):
                cutmix_mixup(input_with_bad_type)

        with pytest.raises(ValueError, match="Could not infer where the labels are"):
            cutmix_mixup({"img": imgs, "Nothing_else": 3})

        with pytest.raises(ValueError, match="labels tensor should be of shape"):
            # Note: the error message isn't ideal, but that's because the label heuristic found the img as the label
            # It's OK, it's an edge-case. The important thing is that this fails loudly instead of passing silently
            cutmix_mixup(imgs)

        with pytest.raises(ValueError, match="When using the default labels_getter"):
            cutmix_mixup(imgs, "not_a_tensor")

        with pytest.raises(ValueError, match="labels tensor should be of shape"):
            cutmix_mixup(imgs, torch.randint(0, 2, size=(2, 3)))

        with pytest.raises(ValueError, match="Expected a batched input with 4 dims"):
            cutmix_mixup(imgs[None, None], torch.randint(0, num_classes, size=(batch_size,)))

        with pytest.raises(ValueError, match="does not match the batch size of the labels"):
            cutmix_mixup(imgs, torch.randint(0, num_classes, size=(batch_size + 1,)))

        with pytest.raises(ValueError, match="labels tensor should be of shape"):
            # The purpose of this check is more about documenting the current
            # behaviour of what happens on a Compose(), rather than actually
            # asserting the expected behaviour. We may support Compose() in the
            # future, e.g. for 2 consecutive CutMix?
            labels = torch.randint(0, num_classes, size=(batch_size,))
            transforms.Compose([cutmix_mixup, cutmix_mixup])(imgs, labels)


@pytest.mark.parametrize("key", ("labels", "LABELS", "LaBeL", "SOME_WEIRD_KEY_THAT_HAS_LABeL_IN_IT"))
@pytest.mark.parametrize("sample_type", (tuple, list, dict))
def test_labels_getter_default_heuristic(key, sample_type):
    labels = torch.arange(10)
    sample = {key: labels, "another_key": "whatever"}
    if sample_type is not dict:
        sample = sample_type((None, sample, "whatever_again"))
    assert transforms._utils._find_labels_default_heuristic(sample) is labels

    if key.lower() != "labels":
        # If "labels" is in the dict (case-insensitive),
        # it takes precedence over other keys which would otherwise be a match
        d = {key: "something_else", "labels": labels}
        assert transforms._utils._find_labels_default_heuristic(d) is labels
2054
2055
2056
2057
2058
2059


class TestShapeGetters:
    @pytest.mark.parametrize(
        ("kernel", "make_input"),
        [
2060
2061
2062
            (F.get_dimensions_image, make_image_tensor),
            (F._get_dimensions_image_pil, make_image_pil),
            (F.get_dimensions_image, make_image),
2063
2064
2065
2066
2067
2068
2069
2070
2071
2072
2073
2074
2075
2076
            (F.get_dimensions_video, make_video),
        ],
    )
    def test_get_dimensions(self, kernel, make_input):
        size = (10, 10)
        color_space, num_channels = "RGB", 3

        input = make_input(size, color_space=color_space)

        assert kernel(input) == F.get_dimensions(input) == [num_channels, *size]

    @pytest.mark.parametrize(
        ("kernel", "make_input"),
        [
2077
2078
2079
            (F.get_num_channels_image, make_image_tensor),
            (F._get_num_channels_image_pil, make_image_pil),
            (F.get_num_channels_image, make_image),
2080
2081
2082
2083
2084
2085
2086
2087
2088
2089
2090
2091
2092
            (F.get_num_channels_video, make_video),
        ],
    )
    def test_get_num_channels(self, kernel, make_input):
        color_space, num_channels = "RGB", 3

        input = make_input(color_space=color_space)

        assert kernel(input) == F.get_num_channels(input) == num_channels

    @pytest.mark.parametrize(
        ("kernel", "make_input"),
        [
2093
2094
2095
            (F.get_size_image, make_image_tensor),
            (F._get_size_image_pil, make_image_pil),
            (F.get_size_image, make_image),
2096
            (F.get_size_bounding_boxes, make_bounding_boxes),
2097
2098
2099
2100
2101
2102
2103
2104
2105
2106
2107
2108
2109
2110
2111
2112
2113
2114
2115
2116
2117
2118
2119
2120
2121
2122
2123
            (F.get_size_mask, make_detection_mask),
            (F.get_size_mask, make_segmentation_mask),
            (F.get_size_video, make_video),
        ],
    )
    def test_get_size(self, kernel, make_input):
        size = (10, 10)

        input = make_input(size)

        assert kernel(input) == F.get_size(input) == list(size)

    @pytest.mark.parametrize(
        ("kernel", "make_input"),
        [
            (F.get_num_frames_video, make_video_tensor),
            (F.get_num_frames_video, make_video),
        ],
    )
    def test_get_num_frames(self, kernel, make_input):
        num_frames = 4

        input = make_input(num_frames=num_frames)

        assert kernel(input) == F.get_num_frames(input) == num_frames

    @pytest.mark.parametrize(
Nicolas Hug's avatar
Nicolas Hug committed
2124
        ("functional", "make_input"),
2125
        [
2126
            (F.get_dimensions, make_bounding_boxes),
2127
2128
            (F.get_dimensions, make_detection_mask),
            (F.get_dimensions, make_segmentation_mask),
2129
            (F.get_num_channels, make_bounding_boxes),
2130
2131
2132
2133
            (F.get_num_channels, make_detection_mask),
            (F.get_num_channels, make_segmentation_mask),
            (F.get_num_frames, make_image_pil),
            (F.get_num_frames, make_image),
2134
            (F.get_num_frames, make_bounding_boxes),
2135
2136
2137
2138
            (F.get_num_frames, make_detection_mask),
            (F.get_num_frames, make_segmentation_mask),
        ],
    )
Nicolas Hug's avatar
Nicolas Hug committed
2139
    def test_unsupported_types(self, functional, make_input):
2140
2141
2142
        input = make_input()

        with pytest.raises(TypeError, match=re.escape(str(type(input)))):
Nicolas Hug's avatar
Nicolas Hug committed
2143
            functional(input)
2144
2145
2146


class TestRegisterKernel:
Nicolas Hug's avatar
Nicolas Hug committed
2147
2148
    @pytest.mark.parametrize("functional", (F.resize, "resize"))
    def test_register_kernel(self, functional):
2149
2150
2151
2152
2153
        class CustomDatapoint(datapoints.Datapoint):
            pass

        kernel_was_called = False

Nicolas Hug's avatar
Nicolas Hug committed
2154
        @F.register_kernel(functional, CustomDatapoint)
2155
2156
2157
2158
2159
2160
2161
2162
2163
2164
2165
2166
2167
2168
2169
2170
        def new_resize(dp, *args, **kwargs):
            nonlocal kernel_was_called
            kernel_was_called = True
            return dp

        t = transforms.Resize(size=(224, 224), antialias=True)

        my_dp = CustomDatapoint(torch.rand(3, 10, 10))
        out = t(my_dp)
        assert out is my_dp
        assert kernel_was_called

        # Sanity check to make sure we didn't override the kernel of other types
        t(torch.rand(3, 10, 10)).shape == (3, 224, 224)
        t(datapoints.Image(torch.rand(3, 10, 10))).shape == (3, 224, 224)

2171
    def test_errors(self):
Nicolas Hug's avatar
Nicolas Hug committed
2172
        with pytest.raises(ValueError, match="Could not find functional with name"):
2173
2174
            F.register_kernel("bad_name", datapoints.Image)

Nicolas Hug's avatar
Nicolas Hug committed
2175
        with pytest.raises(ValueError, match="Kernels can only be registered on functionals"):
2176
2177
2178
2179
2180
            F.register_kernel(datapoints.Image, F.resize)

        with pytest.raises(ValueError, match="Kernels can only be registered for subclasses"):
            F.register_kernel(F.resize, object)

2181
        with pytest.raises(ValueError, match="cannot be registered for the builtin datapoint classes"):
2182
            F.register_kernel(F.resize, datapoints.Image)(F.resize_image)
2183

2184
2185
2186
2187
2188
2189
2190
2191
2192
2193
2194
        class CustomDatapoint(datapoints.Datapoint):
            pass

        def resize_custom_datapoint():
            pass

        F.register_kernel(F.resize, CustomDatapoint)(resize_custom_datapoint)

        with pytest.raises(ValueError, match="already has a kernel registered for type"):
            F.register_kernel(F.resize, CustomDatapoint)(resize_custom_datapoint)

2195
2196

class TestGetKernel:
Nicolas Hug's avatar
Nicolas Hug committed
2197
    # We are using F.resize as functional and the kernels below as proxy. Any other functional / kernels combination
2198
2199
    # would also be fine
    KERNELS = {
2200
2201
2202
        torch.Tensor: F.resize_image,
        PIL.Image.Image: F._resize_image_pil,
        datapoints.Image: F.resize_image,
2203
2204
2205
2206
2207
        datapoints.BoundingBoxes: F.resize_bounding_boxes,
        datapoints.Mask: F.resize_mask,
        datapoints.Video: F.resize_video,
    }

2208
2209
2210
2211
    @pytest.mark.parametrize("input_type", [str, int, object])
    def test_unsupported_types(self, input_type):
        with pytest.raises(TypeError, match="supports inputs of type"):
            _get_kernel(F.resize, input_type)
2212
2213
2214

    def test_exact_match(self):
        # We cannot use F.resize together with self.KERNELS mapping here directly here, since this is only the
Nicolas Hug's avatar
Nicolas Hug committed
2215
        # ideal wrapping. Practically, we have an intermediate wrapper layer. Thus, we create a new resize functional
2216
2217
2218
2219
2220
2221
2222
2223
2224
2225
2226
        # here, register the kernels without wrapper, and check the exact matching afterwards.
        def resize_with_pure_kernels():
            pass

        for input_type, kernel in self.KERNELS.items():
            _register_kernel_internal(resize_with_pure_kernels, input_type, datapoint_wrapper=False)(kernel)

            assert _get_kernel(resize_with_pure_kernels, input_type) is kernel

    def test_builtin_datapoint_subclass(self):
        # We cannot use F.resize together with self.KERNELS mapping here directly here, since this is only the
Nicolas Hug's avatar
Nicolas Hug committed
2227
        # ideal wrapping. Practically, we have an intermediate wrapper layer. Thus, we create a new resize functional
2228
2229
2230
2231
2232
2233
2234
2235
2236
2237
2238
2239
2240
2241
2242
2243
2244
2245
2246
2247
2248
2249
2250
2251
2252
2253
2254
2255
2256
2257
2258
2259
2260
2261
2262
        # here, register the kernels without wrapper, and check if subclasses of our builtin datapoints get dispatched
        # to the kernel of the corresponding superclass
        def resize_with_pure_kernels():
            pass

        class MyImage(datapoints.Image):
            pass

        class MyBoundingBoxes(datapoints.BoundingBoxes):
            pass

        class MyMask(datapoints.Mask):
            pass

        class MyVideo(datapoints.Video):
            pass

        for custom_datapoint_subclass in [
            MyImage,
            MyBoundingBoxes,
            MyMask,
            MyVideo,
        ]:
            builtin_datapoint_class = custom_datapoint_subclass.__mro__[1]
            builtin_datapoint_kernel = self.KERNELS[builtin_datapoint_class]
            _register_kernel_internal(resize_with_pure_kernels, builtin_datapoint_class, datapoint_wrapper=False)(
                builtin_datapoint_kernel
            )

            assert _get_kernel(resize_with_pure_kernels, custom_datapoint_subclass) is builtin_datapoint_kernel

    def test_datapoint_subclass(self):
        class MyDatapoint(datapoints.Datapoint):
            pass

2263
2264
        with pytest.raises(TypeError, match="supports inputs of type"):
            _get_kernel(F.resize, MyDatapoint)
2265
2266
2267
2268
2269
2270
2271

        def resize_my_datapoint():
            pass

        _register_kernel_internal(F.resize, MyDatapoint, datapoint_wrapper=False)(resize_my_datapoint)

        assert _get_kernel(F.resize, MyDatapoint) is resize_my_datapoint
2272

2273
2274
2275
2276
2277
2278
2279
2280
2281
2282
2283
2284
2285
2286
2287
2288
2289
2290
    def test_pil_image_subclass(self):
        opened_image = PIL.Image.open(Path(__file__).parent / "assets" / "encode_jpeg" / "grace_hopper_517x606.jpg")
        loaded_image = opened_image.convert("RGB")

        # check the assumptions
        assert isinstance(opened_image, PIL.Image.Image)
        assert type(opened_image) is not PIL.Image.Image

        assert type(loaded_image) is PIL.Image.Image

        size = [17, 11]
        for image in [opened_image, loaded_image]:
            kernel = _get_kernel(F.resize, type(image))

            output = kernel(image, size=size)

            assert F.get_size(output) == size

2291
2292
2293
2294
2295
2296
2297

class TestPermuteChannels:
    _DEFAULT_PERMUTATION = [2, 0, 1]

    @pytest.mark.parametrize(
        ("kernel", "make_input"),
        [
2298
            (F.permute_channels_image, make_image_tensor),
2299
2300
            # FIXME
            # check_kernel does not support PIL kernel, but it should
2301
            (F.permute_channels_image, make_image),
2302
2303
2304
2305
2306
2307
2308
2309
            (F.permute_channels_video, make_video),
        ],
    )
    @pytest.mark.parametrize("dtype", [torch.float32, torch.uint8])
    @pytest.mark.parametrize("device", cpu_and_cuda())
    def test_kernel(self, kernel, make_input, dtype, device):
        check_kernel(kernel, make_input(dtype=dtype, device=device), permutation=self._DEFAULT_PERMUTATION)

Nicolas Hug's avatar
Nicolas Hug committed
2310
    @pytest.mark.parametrize("make_input", [make_image_tensor, make_image_pil, make_image, make_video])
Nicolas Hug's avatar
Nicolas Hug committed
2311
2312
    def test_functional(self, make_input):
        check_functional(F.permute_channels, make_input(), permutation=self._DEFAULT_PERMUTATION)
2313
2314
2315
2316

    @pytest.mark.parametrize(
        ("kernel", "input_type"),
        [
2317
2318
2319
            (F.permute_channels_image, torch.Tensor),
            (F._permute_channels_image_pil, PIL.Image.Image),
            (F.permute_channels_image, datapoints.Image),
2320
2321
2322
            (F.permute_channels_video, datapoints.Video),
        ],
    )
Nicolas Hug's avatar
Nicolas Hug committed
2323
2324
    def test_functional_signature(self, kernel, input_type):
        check_functional_kernel_signature_match(F.permute_channels, kernel=kernel, input_type=input_type)
2325
2326
2327
2328
2329
2330
2331
2332
2333
2334
2335
2336
2337
2338
2339

    def reference_image_correctness(self, image, permutation):
        channel_images = image.split(1, dim=-3)
        permuted_channel_images = [channel_images[channel_idx] for channel_idx in permutation]
        return datapoints.Image(torch.concat(permuted_channel_images, dim=-3))

    @pytest.mark.parametrize("permutation", [[2, 0, 1], [1, 2, 0], [2, 0, 1], [0, 1, 2]])
    @pytest.mark.parametrize("batch_dims", [(), (2,), (2, 1)])
    def test_image_correctness(self, permutation, batch_dims):
        image = make_image(batch_dims=batch_dims)

        actual = F.permute_channels(image, permutation=permutation)
        expected = self.reference_image_correctness(image, permutation=permutation)

        torch.testing.assert_close(actual, expected)
Philip Meier's avatar
Philip Meier committed
2340
2341
2342
2343
2344
2345
2346
2347
2348
2349
2350
2351
2352
2353
2354
2355
2356
2357
2358
2359
2360
2361


class TestElastic:
    def _make_displacement(self, inpt):
        return torch.rand(
            1,
            *F.get_size(inpt),
            2,
            dtype=torch.float32,
            device=inpt.device if isinstance(inpt, torch.Tensor) else "cpu",
        )

    @param_value_parametrization(
        interpolation=[transforms.InterpolationMode.NEAREST, transforms.InterpolationMode.BILINEAR],
        fill=EXHAUSTIVE_TYPE_FILLS,
    )
    @pytest.mark.parametrize("dtype", [torch.float32, torch.uint8])
    @pytest.mark.parametrize("device", cpu_and_cuda())
    def test_kernel_image_tensor(self, param, value, dtype, device):
        image = make_image_tensor(dtype=dtype, device=device)

        check_kernel(
Philip Meier's avatar
Philip Meier committed
2362
            F.elastic_image,
Philip Meier's avatar
Philip Meier committed
2363
2364
2365
2366
2367
2368
2369
2370
2371
2372
            image,
            displacement=self._make_displacement(image),
            **{param: value},
            check_scripted_vs_eager=not (param == "fill" and isinstance(value, (int, float))),
        )

    @pytest.mark.parametrize("format", list(datapoints.BoundingBoxFormat))
    @pytest.mark.parametrize("dtype", [torch.float32, torch.int64])
    @pytest.mark.parametrize("device", cpu_and_cuda())
    def test_kernel_bounding_boxes(self, format, dtype, device):
2373
        bounding_boxes = make_bounding_boxes(format=format, dtype=dtype, device=device)
Philip Meier's avatar
Philip Meier committed
2374
2375
2376
2377
2378
2379
2380
2381
2382
2383
2384
2385
2386
2387
2388
2389
2390
2391
2392
2393

        check_kernel(
            F.elastic_bounding_boxes,
            bounding_boxes,
            format=bounding_boxes.format,
            canvas_size=bounding_boxes.canvas_size,
            displacement=self._make_displacement(bounding_boxes),
        )

    @pytest.mark.parametrize("make_mask", [make_segmentation_mask, make_detection_mask])
    def test_kernel_mask(self, make_mask):
        mask = make_mask()
        check_kernel(F.elastic_mask, mask, displacement=self._make_displacement(mask))

    def test_kernel_video(self):
        video = make_video()
        check_kernel(F.elastic_video, video, displacement=self._make_displacement(video))

    @pytest.mark.parametrize(
        "make_input",
2394
        [make_image_tensor, make_image_pil, make_image, make_bounding_boxes, make_segmentation_mask, make_video],
Philip Meier's avatar
Philip Meier committed
2395
2396
2397
2398
2399
2400
2401
2402
    )
    def test_functional(self, make_input):
        input = make_input()
        check_functional(F.elastic, input, displacement=self._make_displacement(input))

    @pytest.mark.parametrize(
        ("kernel", "input_type"),
        [
Philip Meier's avatar
Philip Meier committed
2403
2404
2405
            (F.elastic_image, torch.Tensor),
            (F._elastic_image_pil, PIL.Image.Image),
            (F.elastic_image, datapoints.Image),
Philip Meier's avatar
Philip Meier committed
2406
2407
2408
2409
2410
2411
2412
2413
2414
2415
            (F.elastic_bounding_boxes, datapoints.BoundingBoxes),
            (F.elastic_mask, datapoints.Mask),
            (F.elastic_video, datapoints.Video),
        ],
    )
    def test_functional_signature(self, kernel, input_type):
        check_functional_kernel_signature_match(F.elastic, kernel=kernel, input_type=input_type)

    @pytest.mark.parametrize(
        "make_input",
2416
        [make_image_tensor, make_image_pil, make_image, make_bounding_boxes, make_segmentation_mask, make_video],
Philip Meier's avatar
Philip Meier committed
2417
2418
2419
2420
2421
2422
2423
2424
2425
2426
2427
2428
    )
    def test_displacement_error(self, make_input):
        input = make_input()

        with pytest.raises(TypeError, match="displacement should be a Tensor"):
            F.elastic(input, displacement=None)

        with pytest.raises(ValueError, match="displacement shape should be"):
            F.elastic(input, displacement=torch.rand(F.get_size(input)))

    @pytest.mark.parametrize(
        "make_input",
2429
        [make_image_tensor, make_image_pil, make_image, make_bounding_boxes, make_segmentation_mask, make_video],
Philip Meier's avatar
Philip Meier committed
2430
2431
2432
2433
2434
    )
    # ElasticTransform needs larger images to avoid the needed internal padding being larger than the actual image
    @pytest.mark.parametrize("size", [(163, 163), (72, 333), (313, 95)])
    @pytest.mark.parametrize("device", cpu_and_cuda())
    def test_transform(self, make_input, size, device):
2435
2436
2437
2438
2439
2440
        check_transform(
            transforms.ElasticTransform(),
            make_input(size, device=device),
            # We updated gaussian blur kernel generation with a faster and numerically more stable version
            check_v1_compatibility=dict(rtol=0, atol=1),
        )
2441
2442
2443
2444
2445
2446
2447
2448
2449
2450


class TestToPureTensor:
    def test_correctness(self):
        input = {
            "img": make_image(),
            "img_tensor": make_image_tensor(),
            "img_pil": make_image_pil(),
            "mask": make_detection_mask(),
            "video": make_video(),
2451
            "bbox": make_bounding_boxes(),
2452
2453
2454
2455
2456
2457
2458
2459
2460
2461
            "str": "str",
        }

        out = transforms.ToPureTensor()(input)

        for input_value, out_value in zip(input.values(), out.values()):
            if isinstance(input_value, datapoints.Datapoint):
                assert isinstance(out_value, torch.Tensor) and not isinstance(out_value, datapoints.Datapoint)
            else:
                assert isinstance(out_value, type(input_value))