[Enhance] Use Points structure in augmentation and models (#204)

* add h3d backbone

* add h3d backbone

* add h3dnet

* modify scannet config

* fix bugs for proposal refine

* fix bugs for test backbone

* add primitive head test

* modify h3dhead

* modify h3d head

* update loss weight config

* fix bugs for h3d head loss

* modify h3d head get targets function

* update h3dnet base config

* modify weighted loss

* Revert "Merge branch 'h3d_u2' into 'master'"

This reverts merge request !5

* modify pipeline

* modify kitti pipeline

* fix bugs for points rotation

* modify multi sweeps

* modify multi sweep points

* fix bugs for points slicing

* modify BackgroundPointsFilter

* modify pipeline

* modify unittest

* modify unittest

* modify docstring

* modify config files

* update configs

* modify docstring

configs/3dssd/3dssd_kitti-3d-car.py

@@ -27,6 +27,7 @@ file_client_args = dict(backend='disk')
 train_pipeline = [
     dict(
         type='LoadPointsFromFile',
+        coord_type='LIDAR',
         load_dim=4,
         use_dim=4,
         file_client_args=file_client_args),
@@ -58,6 +59,7 @@ train_pipeline = [
 test_pipeline = [
     dict(
         type='LoadPointsFromFile',
+        coord_type='LIDAR',
         load_dim=4,
         use_dim=4,
         file_client_args=file_client_args),

configs/_base_/datasets/kitti-3d-3class.py

@@ -24,6 +24,7 @@ file_client_args = dict(backend='disk')
 train_pipeline = [
     dict(
         type='LoadPointsFromFile',
+        coord_type='LIDAR',
         load_dim=4,
         use_dim=4,
         file_client_args=file_client_args),
@@ -53,6 +54,7 @@ train_pipeline = [
 test_pipeline = [
     dict(
         type='LoadPointsFromFile',
+        coord_type='LIDAR',
         load_dim=4,
         use_dim=4,
         file_client_args=file_client_args),

configs/_base_/datasets/kitti-3d-car.py

@@ -22,6 +22,7 @@ file_client_args = dict(backend='disk')
 train_pipeline = [
     dict(
         type='LoadPointsFromFile',
+        coord_type='LIDAR',
         load_dim=4,
         use_dim=4,
         file_client_args=file_client_args),
@@ -51,6 +52,7 @@ train_pipeline = [
 test_pipeline = [
     dict(
         type='LoadPointsFromFile',
+        coord_type='LIDAR',
         load_dim=4,
         use_dim=4,
         file_client_args=file_client_args),

configs/_base_/datasets/lyft-3d.py

@@ -29,6 +29,7 @@ file_client_args = dict(backend='disk')
 train_pipeline = [
     dict(
         type='LoadPointsFromFile',
+        coord_type='LIDAR',
         load_dim=5,
         use_dim=5,
         file_client_args=file_client_args),
@@ -52,6 +53,7 @@ train_pipeline = [
 test_pipeline = [
     dict(
         type='LoadPointsFromFile',
+        coord_type='LIDAR',
         load_dim=5,
         use_dim=5,
         file_client_args=file_client_args),

configs/_base_/datasets/nus-3d.py

@@ -29,6 +29,7 @@ file_client_args = dict(backend='disk')
 train_pipeline = [
     dict(
         type='LoadPointsFromFile',
+        coord_type='LIDAR',
         load_dim=5,
         use_dim=5,
         file_client_args=file_client_args),
@@ -53,6 +54,7 @@ train_pipeline = [
 test_pipeline = [
     dict(
         type='LoadPointsFromFile',
+        coord_type='LIDAR',
         load_dim=5,
         use_dim=5,
         file_client_args=file_client_args),

configs/_base_/datasets/range100_lyft-3d.py

@@ -29,6 +29,7 @@ file_client_args = dict(backend='disk')
 train_pipeline = [
     dict(
         type='LoadPointsFromFile',
+        coord_type='LIDAR',
         load_dim=5,
         use_dim=5,
         file_client_args=file_client_args),
@@ -52,6 +53,7 @@ train_pipeline = [
 test_pipeline = [
     dict(
         type='LoadPointsFromFile',
+        coord_type='LIDAR',
         load_dim=5,
         use_dim=5,
         file_client_args=file_client_args),

configs/_base_/datasets/scannet-3d-18class.py

@@ -8,6 +8,7 @@ class_names = ('cabinet', 'bed', 'chair', 'sofa', 'table', 'door', 'window',
 train_pipeline = [
     dict(
         type='LoadPointsFromFile',
+        coord_type='DEPTH',
         shift_height=True,
         load_dim=6,
         use_dim=[0, 1, 2]),
@@ -43,6 +44,7 @@ train_pipeline = [
 test_pipeline = [
     dict(
         type='LoadPointsFromFile',
+        coord_type='DEPTH',
         shift_height=True,
         load_dim=6,
         use_dim=[0, 1, 2]),

configs/_base_/datasets/sunrgbd-3d-10class.py

@@ -5,6 +5,7 @@ class_names = ('bed', 'table', 'sofa', 'chair', 'toilet', 'desk', 'dresser',
 train_pipeline = [
     dict(
         type='LoadPointsFromFile',
+        coord_type='DEPTH',
         shift_height=True,
         load_dim=6,
         use_dim=[0, 1, 2]),
@@ -26,6 +27,7 @@ train_pipeline = [
 test_pipeline = [
     dict(
         type='LoadPointsFromFile',
+        coord_type='DEPTH',
         shift_height=True,
         load_dim=6,
         use_dim=[0, 1, 2]),

configs/_base_/datasets/waymoD5-3d-3class.py

@@ -31,6 +31,7 @@ db_sampler = dict(
 train_pipeline = [
     dict(
         type='LoadPointsFromFile',
+        coord_type='LIDAR',
         load_dim=6,
         use_dim=5,
         file_client_args=file_client_args),
@@ -58,6 +59,7 @@ train_pipeline = [
 test_pipeline = [
     dict(
         type='LoadPointsFromFile',
+        coord_type='LIDAR',
         load_dim=6,
         use_dim=5,
         file_client_args=file_client_args),

configs/_base_/datasets/waymoD5-3d-car.py

@@ -22,6 +22,7 @@ db_sampler = dict(
     sample_groups=dict(Car=15),
     points_loader=dict(
         type='LoadPointsFromFile',
+        coord_type='LIDAR',
         load_dim=5,
         use_dim=[0, 1, 2, 3, 4],
         file_client_args=file_client_args))
@@ -29,6 +30,7 @@ db_sampler = dict(
 train_pipeline = [
     dict(
         type='LoadPointsFromFile',
+        coord_type='LIDAR',
         load_dim=6,
         use_dim=5,
         file_client_args=file_client_args),
@@ -56,6 +58,7 @@ train_pipeline = [
 test_pipeline = [
     dict(
         type='LoadPointsFromFile',
+        coord_type='LIDAR',
         load_dim=6,
         use_dim=5,
         file_client_args=file_client_args),

configs/benchmark/hv_PartA2_secfpn_4x8_cyclic_80e_pcdet_kitti-3d-3class.py

@@ -209,7 +209,7 @@ db_sampler = dict(
     classes=class_names,
     sample_groups=dict(Car=20, Pedestrian=15, Cyclist=15))
 train_pipeline = [
-    dict(type='LoadPointsFromFile', load_dim=4, use_dim=4),
+    dict(type='LoadPointsFromFile', coord_type='LIDAR', load_dim=4, use_dim=4),
     dict(type='LoadAnnotations3D', with_bbox_3d=True, with_label_3d=True),
     dict(type='ObjectSample', db_sampler=db_sampler),
     dict(type='RandomFlip3D', flip_ratio_bev_horizontal=0.5),
@@ -225,7 +225,7 @@ train_pipeline = [
     dict(type='Collect3D', keys=['points', 'gt_bboxes_3d', 'gt_labels_3d'])
 ]
 test_pipeline = [
-    dict(type='LoadPointsFromFile', load_dim=4, use_dim=4),
+    dict(type='LoadPointsFromFile', coord_type='LIDAR', load_dim=4, use_dim=4),
     dict(
         type='MultiScaleFlipAug3D',
         img_scale=(1333, 800),

configs/benchmark/hv_pointpillars_secfpn_3x8_100e_det3d_kitti-3d-car.py

@@ -86,7 +86,7 @@ db_sampler = dict(
     classes=class_names)
 
 train_pipeline = [
-    dict(type='LoadPointsFromFile', load_dim=4, use_dim=4),
+    dict(type='LoadPointsFromFile', coord_type='LIDAR', load_dim=4, use_dim=4),
     dict(type='LoadAnnotations3D', with_bbox_3d=True, with_label_3d=True),
     dict(type='ObjectSample', db_sampler=db_sampler),
     dict(
@@ -107,7 +107,7 @@ train_pipeline = [
     dict(type='Collect3D', keys=['points', 'gt_bboxes_3d', 'gt_labels_3d'])
 ]
 test_pipeline = [
-    dict(type='LoadPointsFromFile', load_dim=4, use_dim=4),
+    dict(type='LoadPointsFromFile', coord_type='LIDAR', load_dim=4, use_dim=4),
     dict(type='PointsRangeFilter', point_cloud_range=point_cloud_range),
     dict(
         type='DefaultFormatBundle3D',

configs/benchmark/hv_pointpillars_secfpn_4x8_80e_pcdet_kitti-3d-3class.py

@@ -125,7 +125,7 @@ db_sampler = dict(
     ))
 
 train_pipeline = [
-    dict(type='LoadPointsFromFile', load_dim=4, use_dim=4),
+    dict(type='LoadPointsFromFile', coord_type='LIDAR', load_dim=4, use_dim=4),
     dict(type='LoadAnnotations3D', with_bbox_3d=True, with_label_3d=True),
     dict(type='ObjectSample', db_sampler=db_sampler),
     dict(type='RandomFlip3D', flip_ratio_bev_horizontal=0.5),
@@ -140,7 +140,7 @@ train_pipeline = [
     dict(type='Collect3D', keys=['points', 'gt_bboxes_3d', 'gt_labels_3d']),
 ]
 test_pipeline = [
-    dict(type='LoadPointsFromFile', load_dim=4, use_dim=4),
+    dict(type='LoadPointsFromFile', coord_type='LIDAR', load_dim=4, use_dim=4),
     dict(
         type='MultiScaleFlipAug3D',
         img_scale=(1333, 800),

configs/benchmark/hv_second_secfpn_4x8_80e_pcdet_kitti-3d-3class.py

@@ -119,6 +119,7 @@ file_client_args = dict(backend='disk')
 train_pipeline = [
     dict(
         type='LoadPointsFromFile',
+        coord_type='LIDAR',
         load_dim=4,
         use_dim=4,
         file_client_args=file_client_args),
@@ -142,6 +143,7 @@ train_pipeline = [
 test_pipeline = [
     dict(
         type='LoadPointsFromFile',
+        coord_type='LIDAR',
         load_dim=4,
         use_dim=4,
         file_client_args=file_client_args),

configs/centerpoint/centerpoint_0075voxel_second_secfpn_4x8_cyclic_20e_nus.py

@@ -69,6 +69,7 @@ db_sampler = dict(
 train_pipeline = [
     dict(
         type='LoadPointsFromFile',
+        coord_type='LIDAR',
         load_dim=5,
         use_dim=5,
         file_client_args=file_client_args),
@@ -101,6 +102,7 @@ train_pipeline = [
 test_pipeline = [
     dict(
         type='LoadPointsFromFile',
+        coord_type='LIDAR',
         load_dim=5,
         use_dim=5,
         file_client_args=file_client_args),

configs/centerpoint/centerpoint_0075voxel_second_secfpn_dcn_4x8_cyclic_flip-tta_20e_nus.py

@@ -10,6 +10,7 @@ class_names = [
 test_pipeline = [
     dict(
         type='LoadPointsFromFile',
+        coord_type='LIDAR',
         load_dim=5,
         use_dim=5,
         file_client_args=file_client_args),

configs/centerpoint/centerpoint_0075voxel_second_secfpn_dcn_4x8_cyclic_tta_20e_nus.py

@@ -12,6 +12,7 @@ class_names = [
 test_pipeline = [
     dict(
         type='LoadPointsFromFile',
+        coord_type='LIDAR',
         load_dim=5,
         use_dim=5,
         file_client_args=file_client_args),

configs/centerpoint/centerpoint_0075voxel_second_secfpn_dcn_circlenms_4x8_cyclic_flip-tta_20e_nus.py

@@ -11,6 +11,7 @@ class_names = [
 test_pipeline = [
     dict(
         type='LoadPointsFromFile',
+        coord_type='LIDAR',
         load_dim=5,
         use_dim=5,
         file_client_args=file_client_args),

configs/centerpoint/centerpoint_01voxel_second_secfpn_4x8_cyclic_20e_nus.py

@@ -62,6 +62,7 @@ db_sampler = dict(
 train_pipeline = [
     dict(
         type='LoadPointsFromFile',
+        coord_type='LIDAR',
         load_dim=5,
         use_dim=5,
         file_client_args=file_client_args),
@@ -94,6 +95,7 @@ train_pipeline = [
 test_pipeline = [
     dict(
         type='LoadPointsFromFile',
+        coord_type='LIDAR',
         load_dim=5,
         use_dim=5,
         file_client_args=file_client_args),

configs/centerpoint/centerpoint_02pillar_second_secfpn_4x8_cyclic_20e_nus.py

@@ -63,6 +63,7 @@ db_sampler = dict(
 train_pipeline = [
     dict(
         type='LoadPointsFromFile',
+        coord_type='LIDAR',
         load_dim=5,
         use_dim=5,
         file_client_args=file_client_args),
@@ -95,6 +96,7 @@ train_pipeline = [
 test_pipeline = [
     dict(
         type='LoadPointsFromFile',
+        coord_type='LIDAR',
         load_dim=5,
         use_dim=5,
         file_client_args=file_client_args),