[Question] After I change the point_cloud_range, the runtime is very slow

[Xiangzhaohong]

@skyhehe123 I meet a issue when I change the point_cloud_range for test.

Here is the car_cfg.py:

val=dict( type=dataset_type, root=data_root + 'training/', ann_file=data_root + 'ImageSets/val.txt', img_prefix=None, img_scale=(1242, 375), img_norm_cfg=img_norm_cfg, size_divisor=32, flip_ratio=0, with_mask=False, with_label=False, with_point=True, class_names = ['Car'], generator=dict( type='VoxelGenerator', voxel_size=[0.05, 0.05, 0.1], point_cloud_range=[-40,-35,-3,40,35.4,1],#[0., -40., -3., 70.4, 40., 1.], max_num_points=5, max_voxels=20000 ), anchor_generator=dict( type='AnchorGeneratorStride', sizes=[1.6, 3.9, 1.56], anchor_strides=[0.4, 0.4, 1.0], anchor_offsets=[0.2, -39.8, -1.78], rotations=[0, 1.57], ),

Here is the run time:

root@aff8269f7170:/workspace/SA-SSD# python test.py configs/car_cfg.py epoch_50.pth [40, 1600, 1408] [ ] 1/3769, 0.9 task/s, elapsed: 1s, ETA: 4377stime cost: 0.6650443077087402 [ ] 2/3769, 1.5 task/s, elapsed: 1s, ETA: 2436stime cost: 0.1284785270690918 [ ] 3/3769, 1.5 task/s, elapsed: 2s, ETA: 2539stime cost: 0.7264647483825684 [ ] 4/3769, 1.6 task/s, elapsed: 2s, ETA: 2298stime cost: 0.41581201553344727 [ ] 5/3769, 1.7 task/s, elapsed: 3s, ETA: 2255stime cost: 0.5518581867218018 [ ] 6/3769, 1.9 task/s, elapsed: 3s, ETA: 2032stime cost: 0.24195361137390137 [ ] 7/3769, 2.0 task/s, elapsed: 4s, ETA: 1907stime cost: 0.3057591915130615 [ ] 8/3769, 2.1 task/s, elapsed: 4s, ETA: 1759stime cost: 0.18953967094421387 [ ] 9/3769, 2.3 task/s, elapsed: 4s, ETA: 1662stime cost: 0.2331247329711914 [ ] 10/3769, 2.4 task/s, elapsed: 4s, ETA: 1580stime cost: 0.22430849075317383 [ ] 11/3769, 2.3 task/s, elapsed: 5s, ETA: 1605stime cost: 0.49195098876953125 [ ] 12/3769, 2.3 task/s, elapsed: 5s, ETA: 1600stime cost: 0.41010499000549316 [ ] 13/3769, 2.5 task/s, elapsed: 5s, ETA: 1532stime cost: 0.1868739128112793 [ ] 14/3769, 2.5 task/s, elapsed: 6s, ETA: 1505stime cost: 0.3066904544830322 [ ] 15/3769, 2.6 task/s, elapsed: 6s, ETA: 1467stime cost: 0.24893426895141602 [ ] 16/3769, 2.6 task/s, elapsed: 6s, ETA: 1453stime cost: 0.32833147048950195 [ ] 17/3769, 2.5 task/s, elapsed: 7s, ETA: 1520stime cost: 0.687680721282959 [ ] 18/3769, 2.4 task/s, elapsed: 7s, ETA: 1561stime cost: 0.6009337902069092 [ ] 19/3769, 2.5 task/s, elapsed: 8s, ETA: 1506stime cost: 0.14090251922607422 [ ] 20/3769, 2.6 task/s, elapsed: 8s, ETA: 1467stime cost: 0.19079923629760742 [ ] 21/3769, 2.6 task/s, elapsed: 8s, ETA: 1457stime cost: 0.3329944610595703 [ ] 22/3769, 2.6 task/s, elapsed: 9s, ETA: 1465stime cost: 0.43630194664001465 [ ] 23/3769, 2.6 task/s, elapsed: 9s, ETA: 1440stime cost: 0.24027013778686523 [ ] 24/3769, 2.6 task/s, elapsed: 9s, ETA: 1424stime cost: 0.2764148712158203 [ ] 25/3769, 2.7 task/s, elapsed: 9s, ETA: 1389stime cost: 0.14836907386779785 [ ] 26/3769, 2.6 task/s, elapsed: 10s, ETA: 1421stime cost: 0.5922369956970215 [ ] 27/3769, 2.7 task/s, elapsed: 10s, ETA: 1404stime cost: 0.2609384059906006

Thanks a lot!