transforms_3d.py 50.7 KB
Newer Older
zhangwenwei's avatar
zhangwenwei committed
1
import numpy as np
2
import warnings
3
from mmcv import is_tuple_of
4
from mmcv.utils import build_from_cfg
zhangwenwei's avatar
zhangwenwei committed
5

6
from mmdet3d.core import VoxelGenerator
zhangwenwei's avatar
zhangwenwei committed
7
from mmdet3d.core.bbox import box_np_ops
8
from mmdet.datasets.builder import PIPELINES
zhangwenwei's avatar
zhangwenwei committed
9
from mmdet.datasets.pipelines import RandomFlip
10
from ..builder import OBJECTSAMPLERS
zhangwenwei's avatar
zhangwenwei committed
11
12
13
from .data_augment_utils import noise_per_object_v3_


14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
@PIPELINES.register_module()
class RandomDropPointsColor(object):
    r"""Randomly set the color of points to all zeros.

    Once this transform is executed, all the points' color will be dropped.
    Refer to `PAConv <https://github.com/CVMI-Lab/PAConv/blob/main/scene_seg/
    util/transform.py#L223>`_ for more details.

    Args:
        drop_ratio (float): The probability of dropping point colors.
            Defaults to 0.2.
    """

    def __init__(self, drop_ratio=0.2):
        assert isinstance(drop_ratio, (int, float)) and 0 <= drop_ratio <= 1, \
            f'invalid drop_ratio value {drop_ratio}'
        self.drop_ratio = drop_ratio

    def __call__(self, input_dict):
        """Call function to drop point colors.

        Args:
            input_dict (dict): Result dict from loading pipeline.

        Returns:
            dict: Results after color dropping, \
                'points' key is updated in the result dict.
        """
        points = input_dict['points']
        assert points.attribute_dims is not None and \
            'color' in points.attribute_dims, \
            'Expect points have color attribute'

        if np.random.rand() < self.drop_ratio:
            points.color = points.color * 0.0
        return input_dict

    def __repr__(self):
        """str: Return a string that describes the module."""
        repr_str = self.__class__.__name__
        repr_str += f'(drop_ratio={self.drop_ratio})'
        return repr_str


58
@PIPELINES.register_module()
zhangwenwei's avatar
zhangwenwei committed
59
60
61
62
63
64
65
66
class RandomFlip3D(RandomFlip):
    """Flip the points & bbox.

    If the input dict contains the key "flip", then the flag will be used,
    otherwise it will be randomly decided by a ratio specified in the init
    method.

    Args:
zhangwenwei's avatar
zhangwenwei committed
67
68
69
        sync_2d (bool, optional): Whether to apply flip according to the 2D
            images. If True, it will apply the same flip as that to 2D images.
            If False, it will decide whether to flip randomly and independently
liyinhao's avatar
liyinhao committed
70
            to that of 2D images. Defaults to True.
wuyuefeng's avatar
wuyuefeng committed
71
        flip_ratio_bev_horizontal (float, optional): The flipping probability
liyinhao's avatar
liyinhao committed
72
            in horizontal direction. Defaults to 0.0.
wuyuefeng's avatar
wuyuefeng committed
73
        flip_ratio_bev_vertical (float, optional): The flipping probability
liyinhao's avatar
liyinhao committed
74
            in vertical direction. Defaults to 0.0.
zhangwenwei's avatar
zhangwenwei committed
75
76
    """

wuyuefeng's avatar
wuyuefeng committed
77
78
79
80
81
82
83
    def __init__(self,
                 sync_2d=True,
                 flip_ratio_bev_horizontal=0.0,
                 flip_ratio_bev_vertical=0.0,
                 **kwargs):
        super(RandomFlip3D, self).__init__(
            flip_ratio=flip_ratio_bev_horizontal, **kwargs)
zhangwenwei's avatar
zhangwenwei committed
84
        self.sync_2d = sync_2d
wuyuefeng's avatar
wuyuefeng committed
85
86
87
88
89
90
91
92
93
94
95
        self.flip_ratio_bev_vertical = flip_ratio_bev_vertical
        if flip_ratio_bev_horizontal is not None:
            assert isinstance(
                flip_ratio_bev_horizontal,
                (int, float)) and 0 <= flip_ratio_bev_horizontal <= 1
        if flip_ratio_bev_vertical is not None:
            assert isinstance(
                flip_ratio_bev_vertical,
                (int, float)) and 0 <= flip_ratio_bev_vertical <= 1

    def random_flip_data_3d(self, input_dict, direction='horizontal'):
96
97
98
99
100
101
102
103
104
105
        """Flip 3D data randomly.

        Args:
            input_dict (dict): Result dict from loading pipeline.
            direction (str): Flip direction. Default: horizontal.

        Returns:
            dict: Flipped results, 'points', 'bbox3d_fields' keys are \
                updated in the result dict.
        """
wuyuefeng's avatar
wuyuefeng committed
106
        assert direction in ['horizontal', 'vertical']
107
108
109
110
111
        if len(input_dict['bbox3d_fields']) == 0:  # test mode
            input_dict['bbox3d_fields'].append('empty_box3d')
            input_dict['empty_box3d'] = input_dict['box_type_3d'](
                np.array([], dtype=np.float32))
        assert len(input_dict['bbox3d_fields']) == 1
zhangwenwei's avatar
zhangwenwei committed
112
        for key in input_dict['bbox3d_fields']:
113
114
115
116
117
118
119
120
121
122
123
            if 'points' in input_dict:
                input_dict['points'] = input_dict[key].flip(
                    direction, points=input_dict['points'])
            else:
                input_dict[key].flip(direction)
        if 'centers2d' in input_dict:
            assert self.sync_2d is True and direction == 'horizontal', \
                'Only support sync_2d=True and horizontal flip with images'
            w = input_dict['img_shape'][1]
            input_dict['centers2d'][..., 0] = \
                w - input_dict['centers2d'][..., 0]
zhangwenwei's avatar
zhangwenwei committed
124
125

    def __call__(self, input_dict):
126
127
128
129
130
131
132
133
134
135
136
        """Call function to flip points, values in the ``bbox3d_fields`` and \
        also flip 2D image and its annotations.

        Args:
            input_dict (dict): Result dict from loading pipeline.

        Returns:
            dict: Flipped results, 'flip', 'flip_direction', \
                'pcd_horizontal_flip' and 'pcd_vertical_flip' keys are added \
                into result dict.
        """
zhangwenwei's avatar
zhangwenwei committed
137
        # filp 2D image and its annotations
zhangwenwei's avatar
zhangwenwei committed
138
        super(RandomFlip3D, self).__call__(input_dict)
zhangwenwei's avatar
zhangwenwei committed
139

zhangwenwei's avatar
zhangwenwei committed
140
        if self.sync_2d:
wuyuefeng's avatar
wuyuefeng committed
141
142
            input_dict['pcd_horizontal_flip'] = input_dict['flip']
            input_dict['pcd_vertical_flip'] = False
zhangwenwei's avatar
zhangwenwei committed
143
        else:
wuyuefeng's avatar
wuyuefeng committed
144
145
146
147
148
149
150
151
152
            if 'pcd_horizontal_flip' not in input_dict:
                flip_horizontal = True if np.random.rand(
                ) < self.flip_ratio else False
                input_dict['pcd_horizontal_flip'] = flip_horizontal
            if 'pcd_vertical_flip' not in input_dict:
                flip_vertical = True if np.random.rand(
                ) < self.flip_ratio_bev_vertical else False
                input_dict['pcd_vertical_flip'] = flip_vertical

153
154
155
        if 'transformation_3d_flow' not in input_dict:
            input_dict['transformation_3d_flow'] = []

wuyuefeng's avatar
wuyuefeng committed
156
157
        if input_dict['pcd_horizontal_flip']:
            self.random_flip_data_3d(input_dict, 'horizontal')
158
            input_dict['transformation_3d_flow'].extend(['HF'])
wuyuefeng's avatar
wuyuefeng committed
159
160
        if input_dict['pcd_vertical_flip']:
            self.random_flip_data_3d(input_dict, 'vertical')
161
            input_dict['transformation_3d_flow'].extend(['VF'])
zhangwenwei's avatar
zhangwenwei committed
162
163
        return input_dict

zhangwenwei's avatar
zhangwenwei committed
164
    def __repr__(self):
165
        """str: Return a string that describes the module."""
wuyuefeng's avatar
wuyuefeng committed
166
        repr_str = self.__class__.__name__
167
        repr_str += f'(sync_2d={self.sync_2d},'
168
        repr_str += f' flip_ratio_bev_vertical={self.flip_ratio_bev_vertical})'
wuyuefeng's avatar
wuyuefeng committed
169
        return repr_str
zhangwenwei's avatar
zhangwenwei committed
170

zhangwenwei's avatar
zhangwenwei committed
171

172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
@PIPELINES.register_module()
class RandomJitterPoints(object):
    """Randomly jitter point coordinates.

    Different from the global translation in ``GlobalRotScaleTrans``, here we \
        apply different noises to each point in a scene.

    Args:
        jitter_std (list[float]): The standard deviation of jittering noise.
            This applies random noise to all points in a 3D scene, which is \
            sampled from a gaussian distribution whose standard deviation is \
            set by ``jitter_std``. Defaults to [0.01, 0.01, 0.01]
        clip_range (list[float] | None): Clip the randomly generated jitter \
            noise into this range. If None is given, don't perform clipping.
            Defaults to [-0.05, 0.05]

    Note:
        This transform should only be used in point cloud segmentation tasks \
            because we don't transform ground-truth bboxes accordingly.
        For similar transform in detection task, please refer to `ObjectNoise`.
    """

    def __init__(self,
                 jitter_std=[0.01, 0.01, 0.01],
                 clip_range=[-0.05, 0.05]):
        seq_types = (list, tuple, np.ndarray)
        if not isinstance(jitter_std, seq_types):
            assert isinstance(jitter_std, (int, float)), \
                f'unsupported jitter_std type {type(jitter_std)}'
            jitter_std = [jitter_std, jitter_std, jitter_std]
        self.jitter_std = jitter_std

        if clip_range is not None:
            if not isinstance(clip_range, seq_types):
                assert isinstance(clip_range, (int, float)), \
                    f'unsupported clip_range type {type(clip_range)}'
                clip_range = [-clip_range, clip_range]
        self.clip_range = clip_range

    def __call__(self, input_dict):
        """Call function to jitter all the points in the scene.

        Args:
            input_dict (dict): Result dict from loading pipeline.

        Returns:
            dict: Results after adding noise to each point, \
                'points' key is updated in the result dict.
        """
        points = input_dict['points']
        jitter_std = np.array(self.jitter_std, dtype=np.float32)
        jitter_noise = \
            np.random.randn(points.shape[0], 3) * jitter_std[None, :]
        if self.clip_range is not None:
            jitter_noise = np.clip(jitter_noise, self.clip_range[0],
                                   self.clip_range[1])

        points.translate(jitter_noise)
        return input_dict

    def __repr__(self):
        """str: Return a string that describes the module."""
        repr_str = self.__class__.__name__
        repr_str += f'(jitter_std={self.jitter_std},'
        repr_str += f' clip_range={self.clip_range})'
        return repr_str


240
@PIPELINES.register_module()
zhangwenwei's avatar
zhangwenwei committed
241
class ObjectSample(object):
zhangwenwei's avatar
zhangwenwei committed
242
    """Sample GT objects to the data.
zhangwenwei's avatar
zhangwenwei committed
243
244
245
246
247

    Args:
        db_sampler (dict): Config dict of the database sampler.
        sample_2d (bool): Whether to also paste 2D image patch to the images
            This should be true when applying multi-modality cut-and-paste.
liyinhao's avatar
liyinhao committed
248
            Defaults to False.
zhangwenwei's avatar
zhangwenwei committed
249
    """
zhangwenwei's avatar
zhangwenwei committed
250
251
252
253
254
255
256
257
258
259

    def __init__(self, db_sampler, sample_2d=False):
        self.sampler_cfg = db_sampler
        self.sample_2d = sample_2d
        if 'type' not in db_sampler.keys():
            db_sampler['type'] = 'DataBaseSampler'
        self.db_sampler = build_from_cfg(db_sampler, OBJECTSAMPLERS)

    @staticmethod
    def remove_points_in_boxes(points, boxes):
260
261
262
        """Remove the points in the sampled bounding boxes.

        Args:
263
            points (:obj:`BasePoints`): Input point cloud array.
264
265
266
267
268
            boxes (np.ndarray): Sampled ground truth boxes.

        Returns:
            np.ndarray: Points with those in the boxes removed.
        """
269
        masks = box_np_ops.points_in_rbbox(points.coord.numpy(), boxes)
zhangwenwei's avatar
zhangwenwei committed
270
271
272
273
        points = points[np.logical_not(masks.any(-1))]
        return points

    def __call__(self, input_dict):
274
275
276
277
278
279
280
281
282
283
        """Call function to sample ground truth objects to the data.

        Args:
            input_dict (dict): Result dict from loading pipeline.

        Returns:
            dict: Results after object sampling augmentation, \
                'points', 'gt_bboxes_3d', 'gt_labels_3d' keys are updated \
                in the result dict.
        """
zhangwenwei's avatar
zhangwenwei committed
284
        gt_bboxes_3d = input_dict['gt_bboxes_3d']
zhangwenwei's avatar
zhangwenwei committed
285
286
        gt_labels_3d = input_dict['gt_labels_3d']

zhangwenwei's avatar
zhangwenwei committed
287
288
289
        # change to float for blending operation
        points = input_dict['points']
        if self.sample_2d:
wuyuefeng's avatar
wuyuefeng committed
290
            img = input_dict['img']
zhangwenwei's avatar
zhangwenwei committed
291
292
293
            gt_bboxes_2d = input_dict['gt_bboxes']
            # Assume for now 3D & 2D bboxes are the same
            sampled_dict = self.db_sampler.sample_all(
294
295
296
297
                gt_bboxes_3d.tensor.numpy(),
                gt_labels_3d,
                gt_bboxes_2d=gt_bboxes_2d,
                img=img)
zhangwenwei's avatar
zhangwenwei committed
298
299
        else:
            sampled_dict = self.db_sampler.sample_all(
300
                gt_bboxes_3d.tensor.numpy(), gt_labels_3d, img=None)
zhangwenwei's avatar
zhangwenwei committed
301
302
303
304

        if sampled_dict is not None:
            sampled_gt_bboxes_3d = sampled_dict['gt_bboxes_3d']
            sampled_points = sampled_dict['points']
zhangwenwei's avatar
zhangwenwei committed
305
            sampled_gt_labels = sampled_dict['gt_labels_3d']
zhangwenwei's avatar
zhangwenwei committed
306

zhangwenwei's avatar
zhangwenwei committed
307
308
            gt_labels_3d = np.concatenate([gt_labels_3d, sampled_gt_labels],
                                          axis=0)
309
310
311
            gt_bboxes_3d = gt_bboxes_3d.new_box(
                np.concatenate(
                    [gt_bboxes_3d.tensor.numpy(), sampled_gt_bboxes_3d]))
zhangwenwei's avatar
zhangwenwei committed
312

zhangwenwei's avatar
zhangwenwei committed
313
314
            points = self.remove_points_in_boxes(points, sampled_gt_bboxes_3d)
            # check the points dimension
315
            points = points.cat([sampled_points, points])
zhangwenwei's avatar
zhangwenwei committed
316
317
318
319
320

            if self.sample_2d:
                sampled_gt_bboxes_2d = sampled_dict['gt_bboxes_2d']
                gt_bboxes_2d = np.concatenate(
                    [gt_bboxes_2d, sampled_gt_bboxes_2d]).astype(np.float32)
zhangwenwei's avatar
zhangwenwei committed
321

zhangwenwei's avatar
zhangwenwei committed
322
                input_dict['gt_bboxes'] = gt_bboxes_2d
wuyuefeng's avatar
wuyuefeng committed
323
                input_dict['img'] = sampled_dict['img']
zhangwenwei's avatar
zhangwenwei committed
324
325

        input_dict['gt_bboxes_3d'] = gt_bboxes_3d
326
        input_dict['gt_labels_3d'] = gt_labels_3d.astype(np.long)
zhangwenwei's avatar
zhangwenwei committed
327
        input_dict['points'] = points
zhangwenwei's avatar
zhangwenwei committed
328

zhangwenwei's avatar
zhangwenwei committed
329
330
331
        return input_dict

    def __repr__(self):
332
        """str: Return a string that describes the module."""
333
334
335
336
337
338
339
340
341
        repr_str = self.__class__.__name__
        repr_str += f' sample_2d={self.sample_2d},'
        repr_str += f' data_root={self.sampler_cfg.data_root},'
        repr_str += f' info_path={self.sampler_cfg.info_path},'
        repr_str += f' rate={self.sampler_cfg.rate},'
        repr_str += f' prepare={self.sampler_cfg.prepare},'
        repr_str += f' classes={self.sampler_cfg.classes},'
        repr_str += f' sample_groups={self.sampler_cfg.sample_groups}'
        return repr_str
zhangwenwei's avatar
zhangwenwei committed
342
343


344
@PIPELINES.register_module()
zhangwenwei's avatar
zhangwenwei committed
345
class ObjectNoise(object):
zhangwenwei's avatar
zhangwenwei committed
346
    """Apply noise to each GT objects in the scene.
zhangwenwei's avatar
zhangwenwei committed
347
348

    Args:
349
        translation_std (list[float], optional): Standard deviation of the
zhangwenwei's avatar
zhangwenwei committed
350
351
            distribution where translation noise are sampled from.
            Defaults to [0.25, 0.25, 0.25].
352
        global_rot_range (list[float], optional): Global rotation to the scene.
zhangwenwei's avatar
zhangwenwei committed
353
            Defaults to [0.0, 0.0].
354
        rot_range (list[float], optional): Object rotation range.
zhangwenwei's avatar
zhangwenwei committed
355
356
357
358
            Defaults to [-0.15707963267, 0.15707963267].
        num_try (int, optional): Number of times to try if the noise applied is
            invalid. Defaults to 100.
    """
zhangwenwei's avatar
zhangwenwei committed
359
360

    def __init__(self,
zhangwenwei's avatar
zhangwenwei committed
361
                 translation_std=[0.25, 0.25, 0.25],
zhangwenwei's avatar
zhangwenwei committed
362
                 global_rot_range=[0.0, 0.0],
zhangwenwei's avatar
zhangwenwei committed
363
                 rot_range=[-0.15707963267, 0.15707963267],
zhangwenwei's avatar
zhangwenwei committed
364
                 num_try=100):
zhangwenwei's avatar
zhangwenwei committed
365
        self.translation_std = translation_std
zhangwenwei's avatar
zhangwenwei committed
366
        self.global_rot_range = global_rot_range
zhangwenwei's avatar
zhangwenwei committed
367
        self.rot_range = rot_range
zhangwenwei's avatar
zhangwenwei committed
368
369
370
        self.num_try = num_try

    def __call__(self, input_dict):
371
372
373
374
375
376
377
378
379
        """Call function to apply noise to each ground truth in the scene.

        Args:
            input_dict (dict): Result dict from loading pipeline.

        Returns:
            dict: Results after adding noise to each object, \
                'points', 'gt_bboxes_3d' keys are updated in the result dict.
        """
zhangwenwei's avatar
zhangwenwei committed
380
381
        gt_bboxes_3d = input_dict['gt_bboxes_3d']
        points = input_dict['points']
zhangwenwei's avatar
zhangwenwei committed
382

zhangwenwei's avatar
zhangwenwei committed
383
        # TODO: check this inplace function
384
        numpy_box = gt_bboxes_3d.tensor.numpy()
385
386
        numpy_points = points.tensor.numpy()

zhangwenwei's avatar
zhangwenwei committed
387
        noise_per_object_v3_(
388
            numpy_box,
389
            numpy_points,
zhangwenwei's avatar
zhangwenwei committed
390
391
            rotation_perturb=self.rot_range,
            center_noise_std=self.translation_std,
zhangwenwei's avatar
zhangwenwei committed
392
393
            global_random_rot_range=self.global_rot_range,
            num_try=self.num_try)
394
395

        input_dict['gt_bboxes_3d'] = gt_bboxes_3d.new_box(numpy_box)
396
        input_dict['points'] = points.new_point(numpy_points)
zhangwenwei's avatar
zhangwenwei committed
397
398
399
        return input_dict

    def __repr__(self):
400
        """str: Return a string that describes the module."""
zhangwenwei's avatar
zhangwenwei committed
401
        repr_str = self.__class__.__name__
402
403
404
405
        repr_str += f'(num_try={self.num_try},'
        repr_str += f' translation_std={self.translation_std},'
        repr_str += f' global_rot_range={self.global_rot_range},'
        repr_str += f' rot_range={self.rot_range})'
zhangwenwei's avatar
zhangwenwei committed
406
407
408
        return repr_str


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
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
@PIPELINES.register_module()
class GlobalAlignment(object):
    """Apply global alignment to 3D scene points by rotation and translation.

    Args:
        rotation_axis (int): Rotation axis for points and bboxes rotation.

    Note:
        We do not record the applied rotation and translation as in \
            GlobalRotScaleTrans. Because usually, we do not need to reverse \
            the alignment step.
        For example, ScanNet 3D detection task uses aligned ground-truth \
            bounding boxes for evaluation.
    """

    def __init__(self, rotation_axis):
        self.rotation_axis = rotation_axis

    def _trans_points(self, input_dict, trans_factor):
        """Private function to translate points.

        Args:
            input_dict (dict): Result dict from loading pipeline.
            trans_factor (np.ndarray): Translation vector to be applied.

        Returns:
            dict: Results after translation, 'points' is updated in the dict.
        """
        input_dict['points'].translate(trans_factor)

    def _rot_points(self, input_dict, rot_mat):
        """Private function to rotate bounding boxes and points.

        Args:
            input_dict (dict): Result dict from loading pipeline.
            rot_mat (np.ndarray): Rotation matrix to be applied.

        Returns:
            dict: Results after rotation, 'points' is updated in the dict.
        """
        # input should be rot_mat_T so I transpose it here
        input_dict['points'].rotate(rot_mat.T)

    def _check_rot_mat(self, rot_mat):
        """Check if rotation matrix is valid for self.rotation_axis.

        Args:
            rot_mat (np.ndarray): Rotation matrix to be applied.
        """
        is_valid = np.allclose(np.linalg.det(rot_mat), 1.0)
        valid_array = np.zeros(3)
        valid_array[self.rotation_axis] = 1.0
        is_valid &= (rot_mat[self.rotation_axis, :] == valid_array).all()
        is_valid &= (rot_mat[:, self.rotation_axis] == valid_array).all()
        assert is_valid, f'invalid rotation matrix {rot_mat}'

    def __call__(self, input_dict):
        """Call function to shuffle points.

        Args:
            input_dict (dict): Result dict from loading pipeline.

        Returns:
            dict: Results after global alignment, 'points' and keys in \
                input_dict['bbox3d_fields'] are updated in the result dict.
        """
        assert 'axis_align_matrix' in input_dict['ann_info'].keys(), \
            'axis_align_matrix is not provided in GlobalAlignment'

        axis_align_matrix = input_dict['ann_info']['axis_align_matrix']
        assert axis_align_matrix.shape == (4, 4), \
            f'invalid shape {axis_align_matrix.shape} for axis_align_matrix'
        rot_mat = axis_align_matrix[:3, :3]
        trans_vec = axis_align_matrix[:3, -1]

        self._check_rot_mat(rot_mat)
        self._rot_points(input_dict, rot_mat)
        self._trans_points(input_dict, trans_vec)

        return input_dict

    def __repr__(self):
        repr_str = self.__class__.__name__
        repr_str += f'(rotation_axis={self.rotation_axis})'
        return repr_str


496
@PIPELINES.register_module()
zhangwenwei's avatar
zhangwenwei committed
497
class GlobalRotScaleTrans(object):
zhangwenwei's avatar
zhangwenwei committed
498
    """Apply global rotation, scaling and translation to a 3D scene.
zhangwenwei's avatar
zhangwenwei committed
499
500
501

    Args:
        rot_range (list[float]): Range of rotation angle.
liyinhao's avatar
liyinhao committed
502
            Defaults to [-0.78539816, 0.78539816] (close to [-pi/4, pi/4]).
zhangwenwei's avatar
zhangwenwei committed
503
        scale_ratio_range (list[float]): Range of scale ratio.
liyinhao's avatar
liyinhao committed
504
            Defaults to [0.95, 1.05].
505
506
        translation_std (list[float]): The standard deviation of translation
            noise. This applies random translation to a scene by a noise, which
zhangwenwei's avatar
zhangwenwei committed
507
            is sampled from a gaussian distribution whose standard deviation
liyinhao's avatar
liyinhao committed
508
509
            is set by ``translation_std``. Defaults to [0, 0, 0]
        shift_height (bool): Whether to shift height.
wuyuefeng's avatar
wuyuefeng committed
510
            (the fourth dimension of indoor points) when scaling.
liyinhao's avatar
liyinhao committed
511
            Defaults to False.
zhangwenwei's avatar
zhangwenwei committed
512
    """
zhangwenwei's avatar
zhangwenwei committed
513
514

    def __init__(self,
zhangwenwei's avatar
zhangwenwei committed
515
516
                 rot_range=[-0.78539816, 0.78539816],
                 scale_ratio_range=[0.95, 1.05],
wuyuefeng's avatar
wuyuefeng committed
517
518
                 translation_std=[0, 0, 0],
                 shift_height=False):
519
520
521
522
523
        seq_types = (list, tuple, np.ndarray)
        if not isinstance(rot_range, seq_types):
            assert isinstance(rot_range, (int, float)), \
                f'unsupported rot_range type {type(rot_range)}'
            rot_range = [-rot_range, rot_range]
zhangwenwei's avatar
zhangwenwei committed
524
        self.rot_range = rot_range
525
526
527

        assert isinstance(scale_ratio_range, seq_types), \
            f'unsupported scale_ratio_range type {type(scale_ratio_range)}'
zhangwenwei's avatar
zhangwenwei committed
528
        self.scale_ratio_range = scale_ratio_range
529
530
531
532
533
534
535

        if not isinstance(translation_std, seq_types):
            assert isinstance(translation_std, (int, float)), \
                f'unsupported translation_std type {type(translation_std)}'
            translation_std = [
                translation_std, translation_std, translation_std
            ]
536
537
        assert all([std >= 0 for std in translation_std]), \
            'translation_std should be positive'
zhangwenwei's avatar
zhangwenwei committed
538
        self.translation_std = translation_std
wuyuefeng's avatar
wuyuefeng committed
539
        self.shift_height = shift_height
zhangwenwei's avatar
zhangwenwei committed
540
541

    def _trans_bbox_points(self, input_dict):
542
543
544
545
546
547
548
549
550
551
        """Private function to translate bounding boxes and points.

        Args:
            input_dict (dict): Result dict from loading pipeline.

        Returns:
            dict: Results after translation, 'points', 'pcd_trans' \
                and keys in input_dict['bbox3d_fields'] are updated \
                in the result dict.
        """
552
        translation_std = np.array(self.translation_std, dtype=np.float32)
zhangwenwei's avatar
zhangwenwei committed
553
554
        trans_factor = np.random.normal(scale=translation_std, size=3).T

555
        input_dict['points'].translate(trans_factor)
zhangwenwei's avatar
zhangwenwei committed
556
557
558
559
560
        input_dict['pcd_trans'] = trans_factor
        for key in input_dict['bbox3d_fields']:
            input_dict[key].translate(trans_factor)

    def _rot_bbox_points(self, input_dict):
561
562
563
564
565
566
567
568
569
570
        """Private function to rotate bounding boxes and points.

        Args:
            input_dict (dict): Result dict from loading pipeline.

        Returns:
            dict: Results after rotation, 'points', 'pcd_rotation' \
                and keys in input_dict['bbox3d_fields'] are updated \
                in the result dict.
        """
zhangwenwei's avatar
zhangwenwei committed
571
        rotation = self.rot_range
zhangwenwei's avatar
zhangwenwei committed
572
        noise_rotation = np.random.uniform(rotation[0], rotation[1])
zhangwenwei's avatar
zhangwenwei committed
573

574
575
576
577
578
579
580
        # if no bbox in input_dict, only rotate points
        if len(input_dict['bbox3d_fields']) == 0:
            rot_mat_T = input_dict['points'].rotate(noise_rotation)
            input_dict['pcd_rotation'] = rot_mat_T
            return

        # rotate points with bboxes
zhangwenwei's avatar
zhangwenwei committed
581
        for key in input_dict['bbox3d_fields']:
wuyuefeng's avatar
wuyuefeng committed
582
583
584
585
586
            if len(input_dict[key].tensor) != 0:
                points, rot_mat_T = input_dict[key].rotate(
                    noise_rotation, input_dict['points'])
                input_dict['points'] = points
                input_dict['pcd_rotation'] = rot_mat_T
587

zhangwenwei's avatar
zhangwenwei committed
588
    def _scale_bbox_points(self, input_dict):
589
590
591
592
593
594
595
596
597
        """Private function to scale bounding boxes and points.

        Args:
            input_dict (dict): Result dict from loading pipeline.

        Returns:
            dict: Results after scaling, 'points'and keys in \
                input_dict['bbox3d_fields'] are updated in the result dict.
        """
zhangwenwei's avatar
zhangwenwei committed
598
        scale = input_dict['pcd_scale_factor']
599
600
        points = input_dict['points']
        points.scale(scale)
wuyuefeng's avatar
wuyuefeng committed
601
        if self.shift_height:
602
603
            assert 'height' in points.attribute_dims.keys(), \
                'setting shift_height=True but points have no height attribute'
604
605
            points.tensor[:, points.attribute_dims['height']] *= scale
        input_dict['points'] = points
wuyuefeng's avatar
wuyuefeng committed
606

zhangwenwei's avatar
zhangwenwei committed
607
608
        for key in input_dict['bbox3d_fields']:
            input_dict[key].scale(scale)
zhangwenwei's avatar
zhangwenwei committed
609

zhangwenwei's avatar
zhangwenwei committed
610
    def _random_scale(self, input_dict):
611
612
613
614
615
616
617
618
619
        """Private function to randomly set the scale factor.

        Args:
            input_dict (dict): Result dict from loading pipeline.

        Returns:
            dict: Results after scaling, 'pcd_scale_factor' are updated \
                in the result dict.
        """
zhangwenwei's avatar
zhangwenwei committed
620
621
622
        scale_factor = np.random.uniform(self.scale_ratio_range[0],
                                         self.scale_ratio_range[1])
        input_dict['pcd_scale_factor'] = scale_factor
zhangwenwei's avatar
zhangwenwei committed
623
624

    def __call__(self, input_dict):
625
626
627
628
629
630
631
632
633
634
635
        """Private function to rotate, scale and translate bounding boxes and \
        points.

        Args:
            input_dict (dict): Result dict from loading pipeline.

        Returns:
            dict: Results after scaling, 'points', 'pcd_rotation',
                'pcd_scale_factor', 'pcd_trans' and keys in \
                input_dict['bbox3d_fields'] are updated in the result dict.
        """
636
637
638
        if 'transformation_3d_flow' not in input_dict:
            input_dict['transformation_3d_flow'] = []

zhangwenwei's avatar
zhangwenwei committed
639
        self._rot_bbox_points(input_dict)
zhangwenwei's avatar
zhangwenwei committed
640

zhangwenwei's avatar
zhangwenwei committed
641
642
643
        if 'pcd_scale_factor' not in input_dict:
            self._random_scale(input_dict)
        self._scale_bbox_points(input_dict)
zhangwenwei's avatar
zhangwenwei committed
644

zhangwenwei's avatar
zhangwenwei committed
645
        self._trans_bbox_points(input_dict)
646
647

        input_dict['transformation_3d_flow'].extend(['R', 'S', 'T'])
zhangwenwei's avatar
zhangwenwei committed
648
649
650
        return input_dict

    def __repr__(self):
651
        """str: Return a string that describes the module."""
zhangwenwei's avatar
zhangwenwei committed
652
        repr_str = self.__class__.__name__
653
654
655
656
        repr_str += f'(rot_range={self.rot_range},'
        repr_str += f' scale_ratio_range={self.scale_ratio_range},'
        repr_str += f' translation_std={self.translation_std},'
        repr_str += f' shift_height={self.shift_height})'
zhangwenwei's avatar
zhangwenwei committed
657
658
659
        return repr_str


660
@PIPELINES.register_module()
zhangwenwei's avatar
zhangwenwei committed
661
class PointShuffle(object):
662
    """Shuffle input points."""
zhangwenwei's avatar
zhangwenwei committed
663
664

    def __call__(self, input_dict):
665
666
667
668
669
670
        """Call function to shuffle points.

        Args:
            input_dict (dict): Result dict from loading pipeline.

        Returns:
671
672
            dict: Results after filtering, 'points', 'pts_instance_mask' \
                and 'pts_semantic_mask' keys are updated in the result dict.
673
        """
674
675
676
677
678
679
680
681
682
683
684
685
        idx = input_dict['points'].shuffle()
        idx = idx.numpy()

        pts_instance_mask = input_dict.get('pts_instance_mask', None)
        pts_semantic_mask = input_dict.get('pts_semantic_mask', None)

        if pts_instance_mask is not None:
            input_dict['pts_instance_mask'] = pts_instance_mask[idx]

        if pts_semantic_mask is not None:
            input_dict['pts_semantic_mask'] = pts_semantic_mask[idx]

zhangwenwei's avatar
zhangwenwei committed
686
687
688
689
690
691
        return input_dict

    def __repr__(self):
        return self.__class__.__name__


692
@PIPELINES.register_module()
zhangwenwei's avatar
zhangwenwei committed
693
class ObjectRangeFilter(object):
694
695
696
697
698
    """Filter objects by the range.

    Args:
        point_cloud_range (list[float]): Point cloud range.
    """
zhangwenwei's avatar
zhangwenwei committed
699
700
701
702
703
704

    def __init__(self, point_cloud_range):
        self.pcd_range = np.array(point_cloud_range, dtype=np.float32)
        self.bev_range = self.pcd_range[[0, 1, 3, 4]]

    def __call__(self, input_dict):
705
706
707
708
709
710
711
712
713
        """Call function to filter objects by the range.

        Args:
            input_dict (dict): Result dict from loading pipeline.

        Returns:
            dict: Results after filtering, 'gt_bboxes_3d', 'gt_labels_3d' \
                keys are updated in the result dict.
        """
zhangwenwei's avatar
zhangwenwei committed
714
        gt_bboxes_3d = input_dict['gt_bboxes_3d']
zhangwenwei's avatar
zhangwenwei committed
715
        gt_labels_3d = input_dict['gt_labels_3d']
716
        mask = gt_bboxes_3d.in_range_bev(self.bev_range)
zhangwenwei's avatar
zhangwenwei committed
717
        gt_bboxes_3d = gt_bboxes_3d[mask]
ZwwWayne's avatar
ZwwWayne committed
718
719
720
721
722
        # mask is a torch tensor but gt_labels_3d is still numpy array
        # using mask to index gt_labels_3d will cause bug when
        # len(gt_labels_3d) == 1, where mask=1 will be interpreted
        # as gt_labels_3d[1] and cause out of index error
        gt_labels_3d = gt_labels_3d[mask.numpy().astype(np.bool)]
zhangwenwei's avatar
zhangwenwei committed
723
724

        # limit rad to [-pi, pi]
725
726
        gt_bboxes_3d.limit_yaw(offset=0.5, period=2 * np.pi)
        input_dict['gt_bboxes_3d'] = gt_bboxes_3d
zhangwenwei's avatar
zhangwenwei committed
727
728
        input_dict['gt_labels_3d'] = gt_labels_3d

zhangwenwei's avatar
zhangwenwei committed
729
730
731
        return input_dict

    def __repr__(self):
732
        """str: Return a string that describes the module."""
zhangwenwei's avatar
zhangwenwei committed
733
        repr_str = self.__class__.__name__
734
        repr_str += f'(point_cloud_range={self.pcd_range.tolist()})'
zhangwenwei's avatar
zhangwenwei committed
735
736
737
        return repr_str


738
@PIPELINES.register_module()
zhangwenwei's avatar
zhangwenwei committed
739
class PointsRangeFilter(object):
740
741
742
743
744
    """Filter points by the range.

    Args:
        point_cloud_range (list[float]): Point cloud range.
    """
zhangwenwei's avatar
zhangwenwei committed
745
746

    def __init__(self, point_cloud_range):
747
        self.pcd_range = np.array(point_cloud_range, dtype=np.float32)
zhangwenwei's avatar
zhangwenwei committed
748
749

    def __call__(self, input_dict):
750
751
752
753
754
755
        """Call function to filter points by the range.

        Args:
            input_dict (dict): Result dict from loading pipeline.

        Returns:
756
757
            dict: Results after filtering, 'points', 'pts_instance_mask' \
                and 'pts_semantic_mask' keys are updated in the result dict.
758
        """
zhangwenwei's avatar
zhangwenwei committed
759
        points = input_dict['points']
760
761
        points_mask = points.in_range_3d(self.pcd_range)
        clean_points = points[points_mask]
zhangwenwei's avatar
zhangwenwei committed
762
        input_dict['points'] = clean_points
763
764
765
766
767
768
769
770
771
772
773
        points_mask = points_mask.numpy()

        pts_instance_mask = input_dict.get('pts_instance_mask', None)
        pts_semantic_mask = input_dict.get('pts_semantic_mask', None)

        if pts_instance_mask is not None:
            input_dict['pts_instance_mask'] = pts_instance_mask[points_mask]

        if pts_semantic_mask is not None:
            input_dict['pts_semantic_mask'] = pts_semantic_mask[points_mask]

zhangwenwei's avatar
zhangwenwei committed
774
775
776
        return input_dict

    def __repr__(self):
777
        """str: Return a string that describes the module."""
zhangwenwei's avatar
zhangwenwei committed
778
        repr_str = self.__class__.__name__
779
        repr_str += f'(point_cloud_range={self.pcd_range.tolist()})'
zhangwenwei's avatar
zhangwenwei committed
780
        return repr_str
zhangwenwei's avatar
zhangwenwei committed
781
782
783
784


@PIPELINES.register_module()
class ObjectNameFilter(object):
zhangwenwei's avatar
zhangwenwei committed
785
    """Filter GT objects by their names.
zhangwenwei's avatar
zhangwenwei committed
786
787

    Args:
liyinhao's avatar
liyinhao committed
788
        classes (list[str]): List of class names to be kept for training.
zhangwenwei's avatar
zhangwenwei committed
789
790
791
792
793
794
795
    """

    def __init__(self, classes):
        self.classes = classes
        self.labels = list(range(len(self.classes)))

    def __call__(self, input_dict):
796
797
798
799
800
801
802
803
804
        """Call function to filter objects by their names.

        Args:
            input_dict (dict): Result dict from loading pipeline.

        Returns:
            dict: Results after filtering, 'gt_bboxes_3d', 'gt_labels_3d' \
                keys are updated in the result dict.
        """
zhangwenwei's avatar
zhangwenwei committed
805
806
807
808
809
810
811
812
813
        gt_labels_3d = input_dict['gt_labels_3d']
        gt_bboxes_mask = np.array([n in self.labels for n in gt_labels_3d],
                                  dtype=np.bool_)
        input_dict['gt_bboxes_3d'] = input_dict['gt_bboxes_3d'][gt_bboxes_mask]
        input_dict['gt_labels_3d'] = input_dict['gt_labels_3d'][gt_bboxes_mask]

        return input_dict

    def __repr__(self):
814
        """str: Return a string that describes the module."""
zhangwenwei's avatar
zhangwenwei committed
815
816
817
        repr_str = self.__class__.__name__
        repr_str += f'(classes={self.classes})'
        return repr_str
wuyuefeng's avatar
wuyuefeng committed
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843


@PIPELINES.register_module()
class IndoorPointSample(object):
    """Indoor point sample.

    Sampling data to a certain number.

    Args:
        name (str): Name of the dataset.
        num_points (int): Number of points to be sampled.
    """

    def __init__(self, num_points):
        self.num_points = num_points

    def points_random_sampling(self,
                               points,
                               num_samples,
                               replace=None,
                               return_choices=False):
        """Points random sampling.

        Sample points to a certain number.

        Args:
844
            points (np.ndarray | :obj:`BasePoints`): 3D Points.
wuyuefeng's avatar
wuyuefeng committed
845
846
            num_samples (int): Number of samples to be sampled.
            replace (bool): Whether the sample is with or without replacement.
liyinhao's avatar
liyinhao committed
847
848
            Defaults to None.
            return_choices (bool): Whether return choice. Defaults to False.
wuyuefeng's avatar
wuyuefeng committed
849
850

        Returns:
851
852
            tuple[np.ndarray] | np.ndarray:

853
                - points (np.ndarray | :obj:`BasePoints`): 3D Points.
854
                - choices (np.ndarray, optional): The generated random samples.
wuyuefeng's avatar
wuyuefeng committed
855
856
857
858
859
860
861
862
863
864
865
        """
        if replace is None:
            replace = (points.shape[0] < num_samples)
        choices = np.random.choice(
            points.shape[0], num_samples, replace=replace)
        if return_choices:
            return points[choices], choices
        else:
            return points[choices]

    def __call__(self, results):
866
867
868
869
870
871
872
873
874
        """Call function to sample points to in indoor scenes.

        Args:
            input_dict (dict): Result dict from loading pipeline.

        Returns:
            dict: Results after sampling, 'points', 'pts_instance_mask' \
                and 'pts_semantic_mask' keys are updated in the result dict.
        """
wuyuefeng's avatar
wuyuefeng committed
875
876
877
        points = results['points']
        points, choices = self.points_random_sampling(
            points, self.num_points, return_choices=True)
878
        results['points'] = points
879

wuyuefeng's avatar
wuyuefeng committed
880
881
882
        pts_instance_mask = results.get('pts_instance_mask', None)
        pts_semantic_mask = results.get('pts_semantic_mask', None)

883
        if pts_instance_mask is not None:
wuyuefeng's avatar
wuyuefeng committed
884
885
            pts_instance_mask = pts_instance_mask[choices]
            results['pts_instance_mask'] = pts_instance_mask
886
887
888

        if pts_semantic_mask is not None:
            pts_semantic_mask = pts_semantic_mask[choices]
wuyuefeng's avatar
wuyuefeng committed
889
890
891
892
893
            results['pts_semantic_mask'] = pts_semantic_mask

        return results

    def __repr__(self):
894
        """str: Return a string that describes the module."""
wuyuefeng's avatar
wuyuefeng committed
895
        repr_str = self.__class__.__name__
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
        repr_str += f'(num_points={self.num_points})'
        return repr_str


@PIPELINES.register_module()
class IndoorPatchPointSample(object):
    r"""Indoor point sample within a patch. Modified from `PointNet++ <https://
    github.com/charlesq34/pointnet2/blob/master/scannet/scannet_dataset.py>`_.

    Sampling data to a certain number for semantic segmentation.

    Args:
        num_points (int): Number of points to be sampled.
        block_size (float, optional): Size of a block to sample points from.
            Defaults to 1.5.
        sample_rate (float, optional): Stride used in sliding patch generation.
912
913
914
            This parameter is unused in `IndoorPatchPointSample` and thus has
            been deprecated. We plan to remove it in the future.
            Defaults to None.
915
916
917
918
919
920
921
        ignore_index (int, optional): Label index that won't be used for the
            segmentation task. This is set in PointSegClassMapping as neg_cls.
            Defaults to None.
        use_normalized_coord (bool, optional): Whether to use normalized xyz as
            additional features. Defaults to False.
        num_try (int, optional): Number of times to try if the patch selected
            is invalid. Defaults to 10.
922
923
924
925
926
927
928
929
930
931
932
933
        enlarge_size (float | None, optional): Enlarge the sampled patch to
            [-block_size / 2 - enlarge_size, block_size / 2 + enlarge_size] as
            an augmentation. If None, set it as 0.01. Defaults to 0.2.
        min_unique_num (int | None, optional): Minimum number of unique points
            the sampled patch should contain. If None, use PointNet++'s method
            to judge uniqueness. Defaults to None.

    Note:
        This transform should only be used in the training process of point
            cloud segmentation tasks. For the sliding patch generation and
            inference process in testing, please refer to the `slide_inference`
            function of `EncoderDecoder3D` class.
934
935
936
937
938
    """

    def __init__(self,
                 num_points,
                 block_size=1.5,
939
                 sample_rate=None,
940
941
                 ignore_index=None,
                 use_normalized_coord=False,
942
943
944
                 num_try=10,
                 enlarge_size=0.2,
                 min_unique_num=None):
945
946
947
948
949
        self.num_points = num_points
        self.block_size = block_size
        self.ignore_index = ignore_index
        self.use_normalized_coord = use_normalized_coord
        self.num_try = num_try
950
951
952
953
954
955
956
        self.enlarge_size = enlarge_size if enlarge_size is not None else 0.01
        self.min_unique_num = min_unique_num

        if sample_rate is not None:
            warnings.warn(
                "'sample_rate' has been deprecated and will be removed in "
                'the future. Please remove them from your code.')
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971

    def _input_generation(self, coords, patch_center, coord_max, attributes,
                          attribute_dims, point_type):
        """Generating model input.

        Generate input by subtracting patch center and adding additional \
            features. Currently support colors and normalized xyz as features.

        Args:
            coords (np.ndarray): Sampled 3D Points.
            patch_center (np.ndarray): Center coordinate of the selected patch.
            coord_max (np.ndarray): Max coordinate of all 3D Points.
            attributes (np.ndarray): features of input points.
            attribute_dims (dict): Dictionary to indicate the meaning of extra
                dimension.
972
            point_type (type): class of input points inherited from BasePoints.
973
974

        Returns:
975
            :obj:`BasePoints`: The generated input data.
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
        """
        # subtract patch center, the z dimension is not centered
        centered_coords = coords.copy()
        centered_coords[:, 0] -= patch_center[0]
        centered_coords[:, 1] -= patch_center[1]

        if self.use_normalized_coord:
            normalized_coord = coords / coord_max
            attributes = np.concatenate([attributes, normalized_coord], axis=1)
            if attribute_dims is None:
                attribute_dims = dict()
            attribute_dims.update(
                dict(normalized_coord=[
                    attributes.shape[1], attributes.shape[1] +
                    1, attributes.shape[1] + 2
                ]))

        points = np.concatenate([centered_coords, attributes], axis=1)
        points = point_type(
            points, points_dim=points.shape[1], attribute_dims=attribute_dims)

        return points

    def _patch_points_sampling(self, points, sem_mask, replace=None):
        """Patch points sampling.

        First sample a valid patch.
        Then sample points within that patch to a certain number.

        Args:
1006
            points (:obj:`BasePoints`): 3D Points.
1007
1008
1009
1010
1011
            sem_mask (np.ndarray): semantic segmentation mask for input points.
            replace (bool): Whether the sample is with or without replacement.
                Defaults to None.

        Returns:
1012
            tuple[:obj:`BasePoints`, np.ndarray] | :obj:`BasePoints`:
1013

1014
                - points (:obj:`BasePoints`): 3D Points.
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
                - choices (np.ndarray): The generated random samples.
        """
        coords = points.coord.numpy()
        attributes = points.tensor[:, 3:].numpy()
        attribute_dims = points.attribute_dims
        point_type = type(points)

        coord_max = np.amax(coords, axis=0)
        coord_min = np.amin(coords, axis=0)

1025
        for _ in range(self.num_try):
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
            # random sample a point as patch center
            cur_center = coords[np.random.choice(coords.shape[0])]

            # boundary of a patch
            cur_max = cur_center + np.array(
                [self.block_size / 2.0, self.block_size / 2.0, 0.0])
            cur_min = cur_center - np.array(
                [self.block_size / 2.0, self.block_size / 2.0, 0.0])
            cur_max[2] = coord_max[2]
            cur_min[2] = coord_min[2]
            cur_choice = np.sum(
1037
1038
                (coords >= (cur_min - self.enlarge_size)) *
                (coords <= (cur_max + self.enlarge_size)),
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
                axis=1) == 3

            if not cur_choice.any():  # no points in this patch
                continue

            cur_coords = coords[cur_choice, :]
            cur_sem_mask = sem_mask[cur_choice]

            # two criterion for patch sampling, adopted from PointNet++
            # points within selected patch shoule be scattered separately
            mask = np.sum(
                (cur_coords >= (cur_min - 0.01)) * (cur_coords <=
                                                    (cur_max + 0.01)),
                axis=1) == 3
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066

            if self.min_unique_num is None:
                # use PointNet++'s method as default
                # [31, 31, 62] are just some big values used to transform
                # coords from 3d array to 1d and then check their uniqueness
                # this is used in all the ScanNet code following PointNet++
                vidx = np.ceil(
                    (cur_coords[mask, :] - cur_min) / (cur_max - cur_min) *
                    np.array([31.0, 31.0, 62.0]))
                vidx = np.unique(vidx[:, 0] * 31.0 * 62.0 + vidx[:, 1] * 62.0 +
                                 vidx[:, 2])
                flag1 = len(vidx) / 31.0 / 31.0 / 62.0 >= 0.02
            else:
                flag1 = mask.sum() >= self.min_unique_num
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124

            # selected patch should contain enough annotated points
            if self.ignore_index is None:
                flag2 = True
            else:
                flag2 = np.sum(cur_sem_mask != self.ignore_index) / \
                               len(cur_sem_mask) >= 0.7

            if flag1 and flag2:
                break

        # random sample idx
        if replace is None:
            replace = (cur_sem_mask.shape[0] < self.num_points)
        choices = np.random.choice(
            np.where(cur_choice)[0], self.num_points, replace=replace)

        # construct model input
        points = self._input_generation(coords[choices], cur_center, coord_max,
                                        attributes[choices], attribute_dims,
                                        point_type)

        return points, choices

    def __call__(self, results):
        """Call function to sample points to in indoor scenes.

        Args:
            input_dict (dict): Result dict from loading pipeline.

        Returns:
            dict: Results after sampling, 'points', 'pts_instance_mask' \
                and 'pts_semantic_mask' keys are updated in the result dict.
        """
        points = results['points']

        assert 'pts_semantic_mask' in results.keys(), \
            'semantic mask should be provided in training and evaluation'
        pts_semantic_mask = results['pts_semantic_mask']

        points, choices = self._patch_points_sampling(points,
                                                      pts_semantic_mask)

        results['points'] = points
        results['pts_semantic_mask'] = pts_semantic_mask[choices]
        pts_instance_mask = results.get('pts_instance_mask', None)
        if pts_instance_mask is not None:
            results['pts_instance_mask'] = pts_instance_mask[choices]

        return results

    def __repr__(self):
        """str: Return a string that describes the module."""
        repr_str = self.__class__.__name__
        repr_str += f'(num_points={self.num_points},'
        repr_str += f' block_size={self.block_size},'
        repr_str += f' ignore_index={self.ignore_index},'
        repr_str += f' use_normalized_coord={self.use_normalized_coord},'
1125
1126
1127
        repr_str += f' num_try={self.num_try},'
        repr_str += f' enlarge_size={self.enlarge_size},'
        repr_str += f' min_unique_num={self.min_unique_num})'
wuyuefeng's avatar
wuyuefeng committed
1128
        return repr_str
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156


@PIPELINES.register_module()
class BackgroundPointsFilter(object):
    """Filter background points near the bounding box.

    Args:
        bbox_enlarge_range (tuple[float], float): Bbox enlarge range.
    """

    def __init__(self, bbox_enlarge_range):
        assert (is_tuple_of(bbox_enlarge_range, float)
                and len(bbox_enlarge_range) == 3) \
            or isinstance(bbox_enlarge_range, float), \
            f'Invalid arguments bbox_enlarge_range {bbox_enlarge_range}'

        if isinstance(bbox_enlarge_range, float):
            bbox_enlarge_range = [bbox_enlarge_range] * 3
        self.bbox_enlarge_range = np.array(
            bbox_enlarge_range, dtype=np.float32)[np.newaxis, :]

    def __call__(self, input_dict):
        """Call function to filter points by the range.

        Args:
            input_dict (dict): Result dict from loading pipeline.

        Returns:
1157
1158
            dict: Results after filtering, 'points', 'pts_instance_mask' \
                and 'pts_semantic_mask' keys are updated in the result dict.
1159
1160
1161
1162
        """
        points = input_dict['points']
        gt_bboxes_3d = input_dict['gt_bboxes_3d']

xiliu8006's avatar
xiliu8006 committed
1163
1164
1165
1166
        # avoid groundtruth being modified
        gt_bboxes_3d_np = gt_bboxes_3d.tensor.clone().numpy()
        gt_bboxes_3d_np[:, :3] = gt_bboxes_3d.gravity_center.clone().numpy()

1167
1168
        enlarged_gt_bboxes_3d = gt_bboxes_3d_np.copy()
        enlarged_gt_bboxes_3d[:, 3:6] += self.bbox_enlarge_range
xiliu8006's avatar
xiliu8006 committed
1169
        points_numpy = points.tensor.clone().numpy()
1170
1171
        foreground_masks = box_np_ops.points_in_rbbox(points_numpy,
                                                      gt_bboxes_3d_np)
1172
        enlarge_foreground_masks = box_np_ops.points_in_rbbox(
1173
            points_numpy, enlarged_gt_bboxes_3d)
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
        foreground_masks = foreground_masks.max(1)
        enlarge_foreground_masks = enlarge_foreground_masks.max(1)
        valid_masks = ~np.logical_and(~foreground_masks,
                                      enlarge_foreground_masks)

        input_dict['points'] = points[valid_masks]
        pts_instance_mask = input_dict.get('pts_instance_mask', None)
        if pts_instance_mask is not None:
            input_dict['pts_instance_mask'] = pts_instance_mask[valid_masks]

        pts_semantic_mask = input_dict.get('pts_semantic_mask', None)
        if pts_semantic_mask is not None:
            input_dict['pts_semantic_mask'] = pts_semantic_mask[valid_masks]
        return input_dict

    def __repr__(self):
        """str: Return a string that describes the module."""
        repr_str = self.__class__.__name__
1192
        repr_str += f'(bbox_enlarge_range={self.bbox_enlarge_range.tolist()})'
1193
        return repr_str
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
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


@PIPELINES.register_module()
class VoxelBasedPointSampler(object):
    """Voxel based point sampler.

    Apply voxel sampling to multiple sweep points.

    Args:
        cur_sweep_cfg (dict): Config for sampling current points.
        prev_sweep_cfg (dict): Config for sampling previous points.
        time_dim (int): Index that indicate the time dimention
            for input points.
    """

    def __init__(self, cur_sweep_cfg, prev_sweep_cfg=None, time_dim=3):
        self.cur_voxel_generator = VoxelGenerator(**cur_sweep_cfg)
        self.cur_voxel_num = self.cur_voxel_generator._max_voxels
        self.time_dim = time_dim
        if prev_sweep_cfg is not None:
            assert prev_sweep_cfg['max_num_points'] == \
                cur_sweep_cfg['max_num_points']
            self.prev_voxel_generator = VoxelGenerator(**prev_sweep_cfg)
            self.prev_voxel_num = self.prev_voxel_generator._max_voxels
        else:
            self.prev_voxel_generator = None
            self.prev_voxel_num = 0

    def _sample_points(self, points, sampler, point_dim):
        """Sample points for each points subset.

        Args:
            points (np.ndarray): Points subset to be sampled.
            sampler (VoxelGenerator): Voxel based sampler for
                each points subset.
            point_dim (int): The dimention of each points

        Returns:
            np.ndarray: Sampled points.
        """
        voxels, coors, num_points_per_voxel = sampler.generate(points)
        if voxels.shape[0] < sampler._max_voxels:
            padding_points = np.zeros([
                sampler._max_voxels - voxels.shape[0], sampler._max_num_points,
                point_dim
            ],
                                      dtype=points.dtype)
            padding_points[:] = voxels[0]
            sample_points = np.concatenate([voxels, padding_points], axis=0)
        else:
            sample_points = voxels

        return sample_points

    def __call__(self, results):
        """Call function to sample points from multiple sweeps.

        Args:
            input_dict (dict): Result dict from loading pipeline.

        Returns:
            dict: Results after sampling, 'points', 'pts_instance_mask' \
                and 'pts_semantic_mask' keys are updated in the result dict.
        """
        points = results['points']
        original_dim = points.shape[1]

        # TODO: process instance and semantic mask while _max_num_points
        # is larger than 1
        # Extend points with seg and mask fields
        map_fields2dim = []
        start_dim = original_dim
1266
1267
        points_numpy = points.tensor.numpy()
        extra_channel = [points_numpy]
1268
1269
1270
1271
1272
1273
1274
1275
1276
        for idx, key in enumerate(results['pts_mask_fields']):
            map_fields2dim.append((key, idx + start_dim))
            extra_channel.append(results[key][..., None])

        start_dim += len(results['pts_mask_fields'])
        for idx, key in enumerate(results['pts_seg_fields']):
            map_fields2dim.append((key, idx + start_dim))
            extra_channel.append(results[key][..., None])

1277
        points_numpy = np.concatenate(extra_channel, axis=-1)
1278
1279
1280
1281
1282

        # Split points into two part, current sweep points and
        # previous sweeps points.
        # TODO: support different sampling methods for next sweeps points
        # and previous sweeps points.
1283
1284
1285
        cur_points_flag = (points_numpy[:, self.time_dim] == 0)
        cur_sweep_points = points_numpy[cur_points_flag]
        prev_sweeps_points = points_numpy[~cur_points_flag]
1286
1287
1288
1289
1290
1291
1292
1293
1294
        if prev_sweeps_points.shape[0] == 0:
            prev_sweeps_points = cur_sweep_points

        # Shuffle points before sampling
        np.random.shuffle(cur_sweep_points)
        np.random.shuffle(prev_sweeps_points)

        cur_sweep_points = self._sample_points(cur_sweep_points,
                                               self.cur_voxel_generator,
1295
                                               points_numpy.shape[1])
1296
1297
1298
        if self.prev_voxel_generator is not None:
            prev_sweeps_points = self._sample_points(prev_sweeps_points,
                                                     self.prev_voxel_generator,
1299
                                                     points_numpy.shape[1])
1300

1301
1302
            points_numpy = np.concatenate(
                [cur_sweep_points, prev_sweeps_points], 0)
1303
        else:
1304
            points_numpy = cur_sweep_points
1305
1306

        if self.cur_voxel_generator._max_num_points == 1:
1307
1308
            points_numpy = points_numpy.squeeze(1)
        results['points'] = points.new_point(points_numpy[..., :original_dim])
1309
1310
1311

        # Restore the correspoinding seg and mask fields
        for key, dim_index in map_fields2dim:
1312
            results[key] = points_numpy[..., dim_index]
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335

        return results

    def __repr__(self):
        """str: Return a string that describes the module."""

        def _auto_indent(repr_str, indent):
            repr_str = repr_str.split('\n')
            repr_str = [' ' * indent + t + '\n' for t in repr_str]
            repr_str = ''.join(repr_str)[:-1]
            return repr_str

        repr_str = self.__class__.__name__
        indent = 4
        repr_str += '(\n'
        repr_str += ' ' * indent + f'num_cur_sweep={self.cur_voxel_num},\n'
        repr_str += ' ' * indent + f'num_prev_sweep={self.prev_voxel_num},\n'
        repr_str += ' ' * indent + f'time_dim={self.time_dim},\n'
        repr_str += ' ' * indent + 'cur_voxel_generator=\n'
        repr_str += f'{_auto_indent(repr(self.cur_voxel_generator), 8)},\n'
        repr_str += ' ' * indent + 'prev_voxel_generator=\n'
        repr_str += f'{_auto_indent(repr(self.prev_voxel_generator), 8)})'
        return repr_str