transforms_3d.py 25.6 KB
Newer Older
zhangwenwei's avatar
zhangwenwei committed
1
import numpy as np
2
from mmcv import is_tuple_of
3
from mmcv.utils import build_from_cfg
zhangwenwei's avatar
zhangwenwei committed
4
5

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


12
@PIPELINES.register_module()
zhangwenwei's avatar
zhangwenwei committed
13
14
15
16
17
18
19
20
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
21
22
23
        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
24
            to that of 2D images. Defaults to True.
wuyuefeng's avatar
wuyuefeng committed
25
        flip_ratio_bev_horizontal (float, optional): The flipping probability
liyinhao's avatar
liyinhao committed
26
            in horizontal direction. Defaults to 0.0.
wuyuefeng's avatar
wuyuefeng committed
27
        flip_ratio_bev_vertical (float, optional): The flipping probability
liyinhao's avatar
liyinhao committed
28
            in vertical direction. Defaults to 0.0.
zhangwenwei's avatar
zhangwenwei committed
29
30
    """

wuyuefeng's avatar
wuyuefeng committed
31
32
33
34
35
36
37
    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
38
        self.sync_2d = sync_2d
wuyuefeng's avatar
wuyuefeng committed
39
40
41
42
43
44
45
46
47
48
49
        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'):
50
51
52
53
54
55
56
57
58
59
        """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
60
        assert direction in ['horizontal', 'vertical']
61
62
63
64
65
        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
66
        for key in input_dict['bbox3d_fields']:
wuyuefeng's avatar
wuyuefeng committed
67
68
            input_dict['points'] = input_dict[key].flip(
                direction, points=input_dict['points'])
zhangwenwei's avatar
zhangwenwei committed
69
70

    def __call__(self, input_dict):
71
72
73
74
75
76
77
78
79
80
81
        """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
82
        # filp 2D image and its annotations
zhangwenwei's avatar
zhangwenwei committed
83
        super(RandomFlip3D, self).__call__(input_dict)
zhangwenwei's avatar
zhangwenwei committed
84

zhangwenwei's avatar
zhangwenwei committed
85
        if self.sync_2d:
wuyuefeng's avatar
wuyuefeng committed
86
87
            input_dict['pcd_horizontal_flip'] = input_dict['flip']
            input_dict['pcd_vertical_flip'] = False
zhangwenwei's avatar
zhangwenwei committed
88
        else:
wuyuefeng's avatar
wuyuefeng committed
89
90
91
92
93
94
95
96
97
98
99
100
101
            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

        if input_dict['pcd_horizontal_flip']:
            self.random_flip_data_3d(input_dict, 'horizontal')
        if input_dict['pcd_vertical_flip']:
            self.random_flip_data_3d(input_dict, 'vertical')
zhangwenwei's avatar
zhangwenwei committed
102
103
        return input_dict

zhangwenwei's avatar
zhangwenwei committed
104
    def __repr__(self):
105
        """str: Return a string that describes the module."""
wuyuefeng's avatar
wuyuefeng committed
106
107
        repr_str = self.__class__.__name__
        repr_str += '(sync_2d={},'.format(self.sync_2d)
yinchimaoliang's avatar
yinchimaoliang committed
108
        repr_str += 'flip_ratio_bev_vertical={})'.format(
wuyuefeng's avatar
wuyuefeng committed
109
110
            self.flip_ratio_bev_vertical)
        return repr_str
zhangwenwei's avatar
zhangwenwei committed
111

zhangwenwei's avatar
zhangwenwei committed
112

113
@PIPELINES.register_module()
zhangwenwei's avatar
zhangwenwei committed
114
class ObjectSample(object):
zhangwenwei's avatar
zhangwenwei committed
115
    """Sample GT objects to the data.
zhangwenwei's avatar
zhangwenwei committed
116
117
118
119
120

    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
121
            Defaults to False.
zhangwenwei's avatar
zhangwenwei committed
122
    """
zhangwenwei's avatar
zhangwenwei committed
123
124
125
126
127
128
129
130
131
132

    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):
133
134
135
136
137
138
139
140
141
        """Remove the points in the sampled bounding boxes.

        Args:
            points (np.ndarray): Input point cloud array.
            boxes (np.ndarray): Sampled ground truth boxes.

        Returns:
            np.ndarray: Points with those in the boxes removed.
        """
zhangwenwei's avatar
zhangwenwei committed
142
143
144
145
146
        masks = box_np_ops.points_in_rbbox(points, boxes)
        points = points[np.logical_not(masks.any(-1))]
        return points

    def __call__(self, input_dict):
147
148
149
150
151
152
153
154
155
156
        """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
157
        gt_bboxes_3d = input_dict['gt_bboxes_3d']
zhangwenwei's avatar
zhangwenwei committed
158
159
        gt_labels_3d = input_dict['gt_labels_3d']

zhangwenwei's avatar
zhangwenwei committed
160
161
162
        # change to float for blending operation
        points = input_dict['points']
        if self.sample_2d:
wuyuefeng's avatar
wuyuefeng committed
163
            img = input_dict['img']
zhangwenwei's avatar
zhangwenwei committed
164
165
166
            gt_bboxes_2d = input_dict['gt_bboxes']
            # Assume for now 3D & 2D bboxes are the same
            sampled_dict = self.db_sampler.sample_all(
167
168
169
170
                gt_bboxes_3d.tensor.numpy(),
                gt_labels_3d,
                gt_bboxes_2d=gt_bboxes_2d,
                img=img)
zhangwenwei's avatar
zhangwenwei committed
171
172
        else:
            sampled_dict = self.db_sampler.sample_all(
173
                gt_bboxes_3d.tensor.numpy(), gt_labels_3d, img=None)
zhangwenwei's avatar
zhangwenwei committed
174
175
176
177

        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
178
            sampled_gt_labels = sampled_dict['gt_labels_3d']
zhangwenwei's avatar
zhangwenwei committed
179

zhangwenwei's avatar
zhangwenwei committed
180
181
            gt_labels_3d = np.concatenate([gt_labels_3d, sampled_gt_labels],
                                          axis=0)
182
183
184
            gt_bboxes_3d = gt_bboxes_3d.new_box(
                np.concatenate(
                    [gt_bboxes_3d.tensor.numpy(), sampled_gt_bboxes_3d]))
zhangwenwei's avatar
zhangwenwei committed
185

zhangwenwei's avatar
zhangwenwei committed
186
187
188
189
190
191
192
193
194
195
            points = self.remove_points_in_boxes(points, sampled_gt_bboxes_3d)
            # check the points dimension
            dim_inds = points.shape[-1]
            points = np.concatenate([sampled_points[:, :dim_inds], points],
                                    axis=0)

            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
196

zhangwenwei's avatar
zhangwenwei committed
197
                input_dict['gt_bboxes'] = gt_bboxes_2d
wuyuefeng's avatar
wuyuefeng committed
198
                input_dict['img'] = sampled_dict['img']
zhangwenwei's avatar
zhangwenwei committed
199
200

        input_dict['gt_bboxes_3d'] = gt_bboxes_3d
zhangwenwei's avatar
zhangwenwei committed
201
        input_dict['gt_labels_3d'] = gt_labels_3d
zhangwenwei's avatar
zhangwenwei committed
202
        input_dict['points'] = points
zhangwenwei's avatar
zhangwenwei committed
203

zhangwenwei's avatar
zhangwenwei committed
204
205
206
        return input_dict

    def __repr__(self):
207
        """str: Return a string that describes the module."""
208
209
210
211
212
213
214
215
216
        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
217
218


219
@PIPELINES.register_module()
zhangwenwei's avatar
zhangwenwei committed
220
class ObjectNoise(object):
zhangwenwei's avatar
zhangwenwei committed
221
    """Apply noise to each GT objects in the scene.
zhangwenwei's avatar
zhangwenwei committed
222
223

    Args:
224
        translation_std (list[float], optional): Standard deviation of the
zhangwenwei's avatar
zhangwenwei committed
225
226
            distribution where translation noise are sampled from.
            Defaults to [0.25, 0.25, 0.25].
227
        global_rot_range (list[float], optional): Global rotation to the scene.
zhangwenwei's avatar
zhangwenwei committed
228
            Defaults to [0.0, 0.0].
229
        rot_range (list[float], optional): Object rotation range.
zhangwenwei's avatar
zhangwenwei committed
230
231
232
233
            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
234
235

    def __init__(self,
zhangwenwei's avatar
zhangwenwei committed
236
                 translation_std=[0.25, 0.25, 0.25],
zhangwenwei's avatar
zhangwenwei committed
237
                 global_rot_range=[0.0, 0.0],
zhangwenwei's avatar
zhangwenwei committed
238
                 rot_range=[-0.15707963267, 0.15707963267],
zhangwenwei's avatar
zhangwenwei committed
239
                 num_try=100):
zhangwenwei's avatar
zhangwenwei committed
240
        self.translation_std = translation_std
zhangwenwei's avatar
zhangwenwei committed
241
        self.global_rot_range = global_rot_range
zhangwenwei's avatar
zhangwenwei committed
242
        self.rot_range = rot_range
zhangwenwei's avatar
zhangwenwei committed
243
244
245
        self.num_try = num_try

    def __call__(self, input_dict):
246
247
248
249
250
251
252
253
254
        """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
255
256
        gt_bboxes_3d = input_dict['gt_bboxes_3d']
        points = input_dict['points']
zhangwenwei's avatar
zhangwenwei committed
257

zhangwenwei's avatar
zhangwenwei committed
258
        # TODO: check this inplace function
259
        numpy_box = gt_bboxes_3d.tensor.numpy()
zhangwenwei's avatar
zhangwenwei committed
260
        noise_per_object_v3_(
261
            numpy_box,
zhangwenwei's avatar
zhangwenwei committed
262
            points,
zhangwenwei's avatar
zhangwenwei committed
263
264
            rotation_perturb=self.rot_range,
            center_noise_std=self.translation_std,
zhangwenwei's avatar
zhangwenwei committed
265
266
            global_random_rot_range=self.global_rot_range,
            num_try=self.num_try)
267
268

        input_dict['gt_bboxes_3d'] = gt_bboxes_3d.new_box(numpy_box)
zhangwenwei's avatar
zhangwenwei committed
269
270
271
272
        input_dict['points'] = points
        return input_dict

    def __repr__(self):
273
        """str: Return a string that describes the module."""
zhangwenwei's avatar
zhangwenwei committed
274
275
        repr_str = self.__class__.__name__
        repr_str += '(num_try={},'.format(self.num_try)
zhangwenwei's avatar
zhangwenwei committed
276
        repr_str += ' translation_std={},'.format(self.translation_std)
zhangwenwei's avatar
zhangwenwei committed
277
        repr_str += ' global_rot_range={},'.format(self.global_rot_range)
zhangwenwei's avatar
zhangwenwei committed
278
        repr_str += ' rot_range={})'.format(self.rot_range)
zhangwenwei's avatar
zhangwenwei committed
279
280
281
        return repr_str


282
@PIPELINES.register_module()
zhangwenwei's avatar
zhangwenwei committed
283
class GlobalRotScaleTrans(object):
zhangwenwei's avatar
zhangwenwei committed
284
    """Apply global rotation, scaling and translation to a 3D scene.
zhangwenwei's avatar
zhangwenwei committed
285
286
287

    Args:
        rot_range (list[float]): Range of rotation angle.
liyinhao's avatar
liyinhao committed
288
            Defaults to [-0.78539816, 0.78539816] (close to [-pi/4, pi/4]).
zhangwenwei's avatar
zhangwenwei committed
289
        scale_ratio_range (list[float]): Range of scale ratio.
liyinhao's avatar
liyinhao committed
290
            Defaults to [0.95, 1.05].
zhangwenwei's avatar
zhangwenwei committed
291
292
293
        translation_std (list[float]): The standard deviation of ranslation
            noise. This apply random translation to a scene by a noise, which
            is sampled from a gaussian distribution whose standard deviation
liyinhao's avatar
liyinhao committed
294
295
            is set by ``translation_std``. Defaults to [0, 0, 0]
        shift_height (bool): Whether to shift height.
wuyuefeng's avatar
wuyuefeng committed
296
            (the fourth dimension of indoor points) when scaling.
liyinhao's avatar
liyinhao committed
297
            Defaults to False.
zhangwenwei's avatar
zhangwenwei committed
298
    """
zhangwenwei's avatar
zhangwenwei committed
299
300

    def __init__(self,
zhangwenwei's avatar
zhangwenwei committed
301
302
                 rot_range=[-0.78539816, 0.78539816],
                 scale_ratio_range=[0.95, 1.05],
wuyuefeng's avatar
wuyuefeng committed
303
304
                 translation_std=[0, 0, 0],
                 shift_height=False):
zhangwenwei's avatar
zhangwenwei committed
305
306
307
        self.rot_range = rot_range
        self.scale_ratio_range = scale_ratio_range
        self.translation_std = translation_std
wuyuefeng's avatar
wuyuefeng committed
308
        self.shift_height = shift_height
zhangwenwei's avatar
zhangwenwei committed
309
310

    def _trans_bbox_points(self, input_dict):
311
312
313
314
315
316
317
318
319
320
        """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.
        """
zhangwenwei's avatar
zhangwenwei committed
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
        if not isinstance(self.translation_std, (list, tuple, np.ndarray)):
            translation_std = [
                self.translation_std, self.translation_std,
                self.translation_std
            ]
        else:
            translation_std = self.translation_std
        translation_std = np.array(translation_std, dtype=np.float32)
        trans_factor = np.random.normal(scale=translation_std, size=3).T

        input_dict['points'][:, :3] += trans_factor
        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):
337
338
339
340
341
342
343
344
345
346
        """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
347
        rotation = self.rot_range
zhangwenwei's avatar
zhangwenwei committed
348
349
350
        if not isinstance(rotation, list):
            rotation = [-rotation, rotation]
        noise_rotation = np.random.uniform(rotation[0], rotation[1])
zhangwenwei's avatar
zhangwenwei committed
351
352

        for key in input_dict['bbox3d_fields']:
wuyuefeng's avatar
wuyuefeng committed
353
354
355
356
357
            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
358

zhangwenwei's avatar
zhangwenwei committed
359
    def _scale_bbox_points(self, input_dict):
360
361
362
363
364
365
366
367
368
        """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
369
370
        scale = input_dict['pcd_scale_factor']
        input_dict['points'][:, :3] *= scale
wuyuefeng's avatar
wuyuefeng committed
371
372
373
        if self.shift_height:
            input_dict['points'][:, -1] *= scale

zhangwenwei's avatar
zhangwenwei committed
374
375
        for key in input_dict['bbox3d_fields']:
            input_dict[key].scale(scale)
zhangwenwei's avatar
zhangwenwei committed
376

zhangwenwei's avatar
zhangwenwei committed
377
    def _random_scale(self, input_dict):
378
379
380
381
382
383
384
385
386
        """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
387
388
389
        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
390
391

    def __call__(self, input_dict):
392
393
394
395
396
397
398
399
400
401
402
        """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.
        """
zhangwenwei's avatar
zhangwenwei committed
403
        self._rot_bbox_points(input_dict)
zhangwenwei's avatar
zhangwenwei committed
404

zhangwenwei's avatar
zhangwenwei committed
405
406
407
        if 'pcd_scale_factor' not in input_dict:
            self._random_scale(input_dict)
        self._scale_bbox_points(input_dict)
zhangwenwei's avatar
zhangwenwei committed
408

zhangwenwei's avatar
zhangwenwei committed
409
        self._trans_bbox_points(input_dict)
zhangwenwei's avatar
zhangwenwei committed
410
411
412
        return input_dict

    def __repr__(self):
413
        """str: Return a string that describes the module."""
zhangwenwei's avatar
zhangwenwei committed
414
        repr_str = self.__class__.__name__
zhangwenwei's avatar
zhangwenwei committed
415
416
417
        repr_str += '(rot_range={},'.format(self.rot_range)
        repr_str += ' scale_ratio_range={},'.format(self.scale_ratio_range)
        repr_str += ' translation_std={})'.format(self.translation_std)
wuyuefeng's avatar
wuyuefeng committed
418
        repr_str += ' shift_height={})'.format(self.shift_height)
zhangwenwei's avatar
zhangwenwei committed
419
420
421
        return repr_str


422
@PIPELINES.register_module()
zhangwenwei's avatar
zhangwenwei committed
423
class PointShuffle(object):
424
    """Shuffle input points."""
zhangwenwei's avatar
zhangwenwei committed
425
426

    def __call__(self, input_dict):
427
428
429
430
431
432
433
434
435
        """Call function to shuffle points.

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

        Returns:
            dict: Results after filtering, 'points' keys are updated \
                in the result dict.
        """
zhangwenwei's avatar
zhangwenwei committed
436
437
438
439
440
441
442
        np.random.shuffle(input_dict['points'])
        return input_dict

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


443
@PIPELINES.register_module()
zhangwenwei's avatar
zhangwenwei committed
444
class ObjectRangeFilter(object):
445
446
447
448
449
    """Filter objects by the range.

    Args:
        point_cloud_range (list[float]): Point cloud range.
    """
zhangwenwei's avatar
zhangwenwei committed
450
451
452
453
454
455

    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):
456
457
458
459
460
461
462
463
464
        """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
465
        gt_bboxes_3d = input_dict['gt_bboxes_3d']
zhangwenwei's avatar
zhangwenwei committed
466
        gt_labels_3d = input_dict['gt_labels_3d']
467
        mask = gt_bboxes_3d.in_range_bev(self.bev_range)
zhangwenwei's avatar
zhangwenwei committed
468
        gt_bboxes_3d = gt_bboxes_3d[mask]
ZwwWayne's avatar
ZwwWayne committed
469
470
471
472
473
        # 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
474
475

        # limit rad to [-pi, pi]
476
477
        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
478
479
        input_dict['gt_labels_3d'] = gt_labels_3d

zhangwenwei's avatar
zhangwenwei committed
480
481
482
        return input_dict

    def __repr__(self):
483
        """str: Return a string that describes the module."""
zhangwenwei's avatar
zhangwenwei committed
484
485
486
487
488
        repr_str = self.__class__.__name__
        repr_str += '(point_cloud_range={})'.format(self.pcd_range.tolist())
        return repr_str


489
@PIPELINES.register_module()
zhangwenwei's avatar
zhangwenwei committed
490
class PointsRangeFilter(object):
491
492
493
494
495
    """Filter points by the range.

    Args:
        point_cloud_range (list[float]): Point cloud range.
    """
zhangwenwei's avatar
zhangwenwei committed
496
497
498
499
500
501

    def __init__(self, point_cloud_range):
        self.pcd_range = np.array(
            point_cloud_range, dtype=np.float32)[np.newaxis, :]

    def __call__(self, input_dict):
502
503
504
505
506
507
508
509
510
        """Call function to filter points by the range.

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

        Returns:
            dict: Results after filtering, 'points' keys are updated \
                in the result dict.
        """
zhangwenwei's avatar
zhangwenwei committed
511
512
513
514
515
516
517
518
519
        points = input_dict['points']
        points_mask = ((points[:, :3] >= self.pcd_range[:, :3])
                       & (points[:, :3] < self.pcd_range[:, 3:]))
        points_mask = points_mask[:, 0] & points_mask[:, 1] & points_mask[:, 2]
        clean_points = points[points_mask, :]
        input_dict['points'] = clean_points
        return input_dict

    def __repr__(self):
520
        """str: Return a string that describes the module."""
zhangwenwei's avatar
zhangwenwei committed
521
522
523
        repr_str = self.__class__.__name__
        repr_str += '(point_cloud_range={})'.format(self.pcd_range.tolist())
        return repr_str
zhangwenwei's avatar
zhangwenwei committed
524
525
526
527


@PIPELINES.register_module()
class ObjectNameFilter(object):
zhangwenwei's avatar
zhangwenwei committed
528
    """Filter GT objects by their names.
zhangwenwei's avatar
zhangwenwei committed
529
530

    Args:
liyinhao's avatar
liyinhao committed
531
        classes (list[str]): List of class names to be kept for training.
zhangwenwei's avatar
zhangwenwei committed
532
533
534
535
536
537
538
    """

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

    def __call__(self, input_dict):
539
540
541
542
543
544
545
546
547
        """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
548
549
550
551
552
553
554
555
556
        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):
557
        """str: Return a string that describes the module."""
zhangwenwei's avatar
zhangwenwei committed
558
559
560
        repr_str = self.__class__.__name__
        repr_str += f'(classes={self.classes})'
        return repr_str
wuyuefeng's avatar
wuyuefeng committed
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586


@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:
587
            points (np.ndarray): 3D Points.
wuyuefeng's avatar
wuyuefeng committed
588
589
            num_samples (int): Number of samples to be sampled.
            replace (bool): Whether the sample is with or without replacement.
liyinhao's avatar
liyinhao committed
590
591
            Defaults to None.
            return_choices (bool): Whether return choice. Defaults to False.
wuyuefeng's avatar
wuyuefeng committed
592
593

        Returns:
594
595
596
597
            tuple[np.ndarray] | np.ndarray:

                - points (np.ndarray): 3D Points.
                - choices (np.ndarray, optional): The generated random samples.
wuyuefeng's avatar
wuyuefeng committed
598
599
600
601
602
603
604
605
606
607
608
        """
        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):
609
610
611
612
613
614
615
616
617
        """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
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
        points = results['points']
        points, choices = self.points_random_sampling(
            points, self.num_points, return_choices=True)
        pts_instance_mask = results.get('pts_instance_mask', None)
        pts_semantic_mask = results.get('pts_semantic_mask', None)
        results['points'] = points

        if pts_instance_mask is not None and pts_semantic_mask is not None:
            pts_instance_mask = pts_instance_mask[choices]
            pts_semantic_mask = pts_semantic_mask[choices]
            results['pts_instance_mask'] = pts_instance_mask
            results['pts_semantic_mask'] = pts_semantic_mask

        return results

    def __repr__(self):
634
        """str: Return a string that describes the module."""
wuyuefeng's avatar
wuyuefeng committed
635
636
637
        repr_str = self.__class__.__name__
        repr_str += '(num_points={})'.format(self.num_points)
        return repr_str
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699


@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:
            dict: Results after filtering, 'points' keys are updated \
                in the result dict.
        """
        points = input_dict['points']
        gt_bboxes_3d = input_dict['gt_bboxes_3d']

        gt_bboxes_3d_np = gt_bboxes_3d.tensor.numpy()
        gt_bboxes_3d_np[:, :3] = gt_bboxes_3d.gravity_center.numpy()
        enlarged_gt_bboxes_3d = gt_bboxes_3d_np.copy()
        enlarged_gt_bboxes_3d[:, 3:6] += self.bbox_enlarge_range
        foreground_masks = box_np_ops.points_in_rbbox(points, gt_bboxes_3d_np)
        enlarge_foreground_masks = box_np_ops.points_in_rbbox(
            points, enlarged_gt_bboxes_3d)
        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__
        repr_str += '(bbox_enlarge_range={})'.format(
            self.bbox_enlarge_range.tolist())
        return repr_str