[Refactor]: remove legacy builder.py and update corresponding unit tests (#2161)

* fix circle ci

* delete model builder

* fix comments

mmdet3d/datasets/__init__.py

 # Copyright (c) OpenMMLab. All rights reserved.
-from .builder import DATASETS, PIPELINES, build_dataset
 from .dataset_wrappers import CBGSDataset
 from .det3d_dataset import Det3DDataset
 from .kitti_dataset import KittiDataset
@@ -27,8 +26,7 @@ from .utils import get_loading_pipeline
 from .waymo_dataset import WaymoDataset
 
 __all__ = [
-    'KittiDataset', 'DATASETS', 'CBGSDataset',
-    'build_dataset', 'NuScenesDataset', 'LyftDataset',
+    'KittiDataset', 'CBGSDataset', 'NuScenesDataset', 'LyftDataset',
     'ObjectSample', 'RandomFlip3D', 'ObjectNoise', 'GlobalRotScaleTrans',
     'PointShuffle', 'ObjectRangeFilter', 'PointsRangeFilter',
     'LoadPointsFromFile', 'S3DISSegDataset', 'S3DISDataset',
@@ -39,6 +37,5 @@ __all__ = [
     'LoadPointsFromMultiSweeps', 'WaymoDataset', 'BackgroundPointsFilter',
     'VoxelBasedPointSampler', 'get_loading_pipeline', 'RandomDropPointsColor',
     'RandomJitterPoints', 'ObjectNameFilter', 'AffineResize',
-    'RandomShiftScale', 'LoadPointsFromDict', 'PIPELINES',
-    'Resize3D', 'RandomResize3D',
+    'RandomShiftScale', 'LoadPointsFromDict', 'Resize3D', 'RandomResize3D',
 ]

mmdet3d/datasets/builder.py

-# Copyright (c) OpenMMLab. All rights reserved.
-import platform
-
-from mmdet.datasets.builder import _concat_dataset
-
-from mmdet3d.registry import DATASETS, TRANSFORMS
-
-if platform.system() != 'Windows':
-    # https://github.com/pytorch/pytorch/issues/973
-    import resource
-    rlimit = resource.getrlimit(resource.RLIMIT_NOFILE)
-    base_soft_limit = rlimit[0]
-    hard_limit = rlimit[1]
-    soft_limit = min(max(4096, base_soft_limit), hard_limit)
-    resource.setrlimit(resource.RLIMIT_NOFILE, (soft_limit, hard_limit))
-
-OBJECTSAMPLERS = TRANSFORMS
-PIPELINES = TRANSFORMS
-
-
-def build_dataset(cfg, default_args=None):
-    from mmengine.dataset import (ClassBalancedDataset, ConcatDataset,
-                                  RepeatDataset)
-
-    from mmdet3d.datasets.dataset_wrappers import CBGSDataset
-    if isinstance(cfg, (list, tuple)):
-        dataset = ConcatDataset([build_dataset(c, default_args) for c in cfg])
-    elif cfg['type'] == 'ConcatDataset':
-        dataset = ConcatDataset(
-            [build_dataset(c, default_args) for c in cfg['datasets']],
-            cfg.get('separate_eval', True))
-    elif cfg['type'] == 'RepeatDataset':
-        dataset = RepeatDataset(
-            build_dataset(cfg['dataset'], default_args), cfg['times'])
-    elif cfg['type'] == 'ClassBalancedDataset':
-        dataset = ClassBalancedDataset(
-            build_dataset(cfg['dataset'], default_args), cfg['oversample_thr'])
-    elif cfg['type'] == 'CBGSDataset':
-        dataset = CBGSDataset(build_dataset(cfg['dataset'], default_args))
-    elif isinstance(cfg.get('ann_file'), (list, tuple)):
-        dataset = _concat_dataset(cfg, default_args)
-    else:
-        dataset = DATASETS.build(cfg, default_args=default_args)
-
-    return dataset

mmdet3d/datasets/det3d_dataset.py

@@ -11,7 +11,7 @@ from mmengine.dataset import BaseDataset
 from mmengine.logging import print_log
 from terminaltables import AsciiTable
 
-from mmdet3d.datasets import DATASETS
+from mmdet3d.registry import DATASETS
 from mmdet3d.structures import get_box_type
 
 

mmdet3d/datasets/kitti_dataset.py

@@ -3,7 +3,7 @@ from typing import Callable, List, Union
 
 import numpy as np
 
-from mmdet3d.datasets import DATASETS
+from mmdet3d.registry import DATASETS
 from mmdet3d.structures import CameraInstance3DBoxes
 from .det3d_dataset import Det3DDataset
 

mmdet3d/models/__init__.py

 # Copyright (c) OpenMMLab. All rights reserved.
 from mmdet3d.models.layers.fusion_layers import *  # noqa: F401,F403
 from .backbones import *  # noqa: F401,F403
-from .builder import (BACKBONES, DETECTORS, FUSION_LAYERS, HEADS, LOSSES,
-                      MIDDLE_ENCODERS, NECKS, ROI_EXTRACTORS, SEGMENTORS,
-                      SHARED_HEADS, VOXEL_ENCODERS, build_backbone,
-                      build_detector, build_fusion_layer, build_head,
-                      build_loss, build_middle_encoder, build_model,
-                      build_neck, build_roi_extractor, build_shared_head,
-                      build_voxel_encoder)
 from .data_preprocessors import *  # noqa: F401,F403
 from .decode_heads import *  # noqa: F401,F403
 from .dense_heads import *  # noqa: F401,F403
@@ -21,12 +14,3 @@ from .segmentors import *  # noqa: F401,F403
 from .test_time_augs import *  # noqa: F401,F403
 from .utils import *  # noqa: F401,F403
 from .voxel_encoders import *  # noqa: F401,F403
-
-__all__ = [
-    'BACKBONES', 'NECKS', 'ROI_EXTRACTORS', 'SHARED_HEADS', 'HEADS', 'LOSSES',
-    'DETECTORS', 'SEGMENTORS', 'VOXEL_ENCODERS', 'MIDDLE_ENCODERS',
-    'FUSION_LAYERS', 'build_backbone', 'build_neck', 'build_roi_extractor',
-    'build_shared_head', 'build_head', 'build_loss', 'build_detector',
-    'build_fusion_layer', 'build_model', 'build_middle_encoder',
-    'build_voxel_encoder'
-]

mmdet3d/models/backbones/multi_backbone.py

@@ -7,7 +7,6 @@ from mmcv.cnn import ConvModule
 from mmengine.model import BaseModule
 from torch import nn as nn
 
-from mmdet3d.models.builder import build_backbone
 from mmdet3d.registry import MODELS
 
 
@@ -57,7 +56,7 @@ class MultiBackbone(BaseModule):
 
         for backbone_cfg in backbones:
             out_channels += backbone_cfg['fp_channels'][-1][-1]
-            self.backbone_list.append(build_backbone(backbone_cfg))
+            self.backbone_list.append(MODELS.build(backbone_cfg))
 
         # Feature aggregation layers
         if aggregation_mlp_channels is None:

mmdet3d/models/builder.py

-# Copyright (c) OpenMMLab. All rights reserved.
-import warnings
-
-from mmdet3d.registry import MODELS
-
-BACKBONES = MODELS
-NECKS = MODELS
-ROI_EXTRACTORS = MODELS
-SHARED_HEADS = MODELS
-HEADS = MODELS
-LOSSES = MODELS
-DETECTORS = MODELS
-SEGMENTORS = MODELS
-VOXEL_ENCODERS = MODELS
-MIDDLE_ENCODERS = MODELS
-FUSION_LAYERS = MODELS
-SEGMENTORS = MODELS
-
-
-def build_backbone(cfg):
-    """Build backbone."""
-    warnings.warn('``build_backbone`` would be deprecated soon, please use '
-                  '``mmdet3d.registry.MODELS.build()`` ')
-
-    return BACKBONES.build(cfg)
-
-
-def build_neck(cfg):
-    """Build neck."""
-    warnings.warn('``build_neck`` would be deprecated soon, please use '
-                  '``mmdet3d.registry.MODELS.build()`` ')
-
-    return NECKS.build(cfg)
-
-
-def build_roi_extractor(cfg):
-    """Build roi extractor."""
-    warnings.warn(
-        '``build_roi_extractor`` would be deprecated soon, please use '
-        '``mmdet3d.registry.MODELS.build()`` ')
-    return ROI_EXTRACTORS.build(cfg)
-
-
-def build_shared_head(cfg):
-    """Build shared head."""
-    warnings.warn('``build_shared_head`` would be deprecated soon, please use '
-                  '``mmdet3d.registry.MODELS.build()`` ')
-    return SHARED_HEADS.build(cfg)
-
-
-def build_head(cfg):
-    """Build head."""
-    warnings.warn('``build_head`` would be deprecated soon, please use '
-                  '``mmdet3d.registry.MODELS.build()`` ')
-    return HEADS.build(cfg)
-
-
-def build_loss(cfg):
-    """Build loss."""
-    warnings.warn('``build_loss`` would be deprecated soon, please use '
-                  '``mmdet3d.registry.MODELS.build()`` ')
-    return LOSSES.build(cfg)
-
-
-def build_detector(cfg, train_cfg=None, test_cfg=None):
-    """Build detector."""
-    if train_cfg is not None or test_cfg is not None:
-        warnings.warn(
-            'train_cfg and test_cfg is deprecated, '
-            'please specify them in model', UserWarning)
-    assert cfg.get('train_cfg') is None or train_cfg is None, \
-        'train_cfg specified in both outer field and model field '
-    assert cfg.get('test_cfg') is None or test_cfg is None, \
-        'test_cfg specified in both outer field and model field '
-    if cfg['type'] in DETECTORS._module_dict.keys():
-        return DETECTORS.build(
-            cfg, default_args=dict(train_cfg=train_cfg, test_cfg=test_cfg))
-
-
-def build_segmentor(cfg, train_cfg=None, test_cfg=None):
-    """Build segmentor."""
-    if train_cfg is not None or test_cfg is not None:
-        warnings.warn(
-            'train_cfg and test_cfg is deprecated, '
-            'please specify them in model', UserWarning)
-    assert cfg.get('train_cfg') is None or train_cfg is None, \
-        'train_cfg specified in both outer field and model field '
-    assert cfg.get('test_cfg') is None or test_cfg is None, \
-        'test_cfg specified in both outer field and model field '
-    return SEGMENTORS.build(
-        cfg, default_args=dict(train_cfg=train_cfg, test_cfg=test_cfg))
-
-
-def build_model(cfg, train_cfg=None, test_cfg=None):
-    """A function wrapper for building 3D detector or segmentor according to
-    cfg.
-
-    Should be deprecated in the future.
-    """
-    if cfg.type in ['EncoderDecoder3D']:
-        return build_segmentor(cfg, train_cfg=train_cfg, test_cfg=test_cfg)
-    else:
-        return build_detector(cfg, train_cfg=train_cfg, test_cfg=test_cfg)
-
-
-def build_voxel_encoder(cfg):
-    """Build voxel encoder."""
-    warnings.warn('``build_voxel_encoder`` would be deprecated soon, please '
-                  'use ``mmdet3d.registry.MODELS.build()`` ')
-    return VOXEL_ENCODERS.build(cfg)
-
-
-def build_middle_encoder(cfg):
-    """Build middle level encoder."""
-    warnings.warn('``build_middle_encoder`` would be deprecated soon, please '
-                  'use ``mmdet3d.registry.MODELS.build()`` ')
-    return MIDDLE_ENCODERS.build(cfg)
-
-
-def build_fusion_layer(cfg):
-    """Build fusion layer."""
-    warnings.warn('``build_fusion_layer`` would be deprecated soon, please '
-                  'use ``mmdet3d.registry.MODELS.build()`` ')
-    return FUSION_LAYERS.build(cfg)

mmdet3d/models/dense_heads/centerpoint_head.py

@@ -13,7 +13,6 @@ from mmdet3d.models.utils import (clip_sigmoid, draw_heatmap_gaussian,
                                   gaussian_radius)
 from mmdet3d.registry import MODELS, TASK_UTILS
 from mmdet3d.structures import Det3DDataSample, xywhr2xyxyr
-from .. import builder
 from ..layers import circle_nms, nms_bev
 
 
@@ -337,7 +336,7 @@ class CenterHead(BaseModule):
             heads.update(dict(heatmap=(num_cls, num_heatmap_convs)))
             separate_head.update(
                 in_channels=share_conv_channel, heads=heads, num_cls=num_cls)
-            self.task_heads.append(builder.build_head(separate_head))
+            self.task_heads.append(MODELS.build(separate_head))
 
     def forward_single(self, x: Tensor) -> dict:
         """Forward function for CenterPoint.

mmdet3d/models/dense_heads/shape_aware_head.py

@@ -15,7 +15,6 @@ from mmdet3d.models.layers import box3d_multiclass_nms
 from mmdet3d.registry import MODELS
 from mmdet3d.structures import limit_period, xywhr2xyxyr
 from mmdet3d.utils import InstanceList, OptInstanceList
-from ..builder import build_head
 from .anchor3d_head import Anchor3DHead
 
 
@@ -220,7 +219,7 @@ class ShapeAwareHead(Anchor3DHead):
                 in_channels=self.in_channels,
                 shared_conv_channels=task['shared_conv_channels'],
                 shared_conv_strides=task['shared_conv_strides'])
-            self.heads.append(build_head(branch))
+            self.heads.append(MODELS.build(branch))
             cls_ptr += task['num_class']
 
     def forward_single(self, x: Tensor) -> Tuple[Tensor]:

mmdet3d/models/detectors/dfm.py

@@ -5,7 +5,6 @@ from mmdet.models.detectors import BaseDetector
 from mmdet3d.registry import MODELS
 from mmdet3d.structures.ops import bbox3d2result
 from mmdet3d.utils import ConfigType
-from ..builder import build_backbone, build_head, build_neck
 
 
 @MODELS.register_module()
@@ -55,34 +54,34 @@ class DfM(BaseDetector):
                  pretrained=None,
                  init_cfg=None):
         super().__init__(init_cfg=init_cfg)
-        self.backbone = build_backbone(backbone)
-        self.neck = build_neck(neck)
+        self.backbone = MODELS.build(backbone)
+        self.neck = MODELS.build(neck)
         if backbone_stereo is not None:
             backbone_stereo.update(cat_img_feature=self.neck.cat_img_feature)
             backbone_stereo.update(in_sem_channels=self.neck.sem_channels[-1])
-            self.backbone_stereo = build_backbone(backbone_stereo)
+            self.backbone_stereo = MODELS.build(backbone_stereo)
             assert self.neck.cat_img_feature == \
                 self.backbone_stereo.cat_img_feature
             assert self.neck.sem_channels[
                 -1] == self.backbone_stereo.in_sem_channels
         if backbone_3d is not None:
-            self.backbone_3d = build_backbone(backbone_3d)
+            self.backbone_3d = MODELS.build(backbone_3d)
         if neck_3d is not None:
-            self.neck_3d = build_neck(neck_3d)
+            self.neck_3d = MODELS.build(neck_3d)
         if neck_2d is not None:
-            self.neck_2d = build_neck(neck_2d)
+            self.neck_2d = MODELS.build(neck_2d)
         if bbox_head_2d is not None:
-            self.bbox_head_2d = build_head(bbox_head_2d)
+            self.bbox_head_2d = MODELS.build(bbox_head_2d)
         if depth_head_2d is not None:
-            self.depth_head_2d = build_head(depth_head_2d)
+            self.depth_head_2d = MODELS.build(depth_head_2d)
         if depth_head is not None:
-            self.depth_head = build_head(depth_head)
+            self.depth_head = MODELS.build(depth_head)
             self.depth_samples = self.depth_head.depth_samples
         self.train_cfg = train_cfg
         self.test_cfg = test_cfg
         bbox_head_3d.update(train_cfg=train_cfg)
         bbox_head_3d.update(test_cfg=test_cfg)
-        self.bbox_head_3d = build_head(bbox_head_3d)
+        self.bbox_head_3d = MODELS.build(bbox_head_3d)
 
     @property
     def with_backbone_3d(self):

mmdet3d/models/roi_heads/bbox_heads/pv_rcnn_bbox_head.py

@@ -10,7 +10,6 @@ from mmengine.model import BaseModule
 from mmengine.structures import InstanceData
 from torch import nn as nn
 
-from mmdet3d.models.builder import build_loss
 from mmdet3d.models.layers import nms_bev, nms_normal_bev
 from mmdet3d.registry import MODELS, TASK_UTILS
 from mmdet3d.structures.bbox_3d import (LiDARInstance3DBoxes,
@@ -76,8 +75,8 @@ class PVRCNNBBoxHead(BaseModule):
         self.with_corner_loss = with_corner_loss
         self.class_agnostic = class_agnostic
         self.bbox_coder = TASK_UTILS.build(bbox_coder)
-        self.loss_bbox = build_loss(loss_bbox)
-        self.loss_cls = build_loss(loss_cls)
+        self.loss_bbox = MODELS.build(loss_bbox)
+        self.loss_cls = MODELS.build(loss_cls)
         self.use_sigmoid_cls = loss_cls.get('use_sigmoid', False)
 
         cls_out_channels = 1 if class_agnostic else num_classes

mmdet3d/models/roi_heads/mask_heads/foreground_segmentation_head.py

@@ -8,7 +8,6 @@ from mmengine.model import BaseModule
 from mmengine.structures import InstanceData
 from torch import nn as nn
 
-from mmdet3d.models.builder import build_loss
 from mmdet3d.registry import MODELS
 from mmdet3d.utils import InstanceList
 
@@ -72,7 +71,7 @@ class ForegroundSegmentationHead(BaseModule):
 
         self.seg_cls_layer = nn.Sequential(*mlps_layers)
 
-        self.loss_seg = build_loss(loss_seg)
+        self.loss_seg = MODELS.build(loss_seg)
 
     def forward(self, feats: torch.Tensor) -> dict:
         """Forward head.

mmdet3d/models/voxel_encoders/voxel_encoder.py

@@ -5,7 +5,6 @@ from mmcv.ops import DynamicScatter
 from torch import Tensor, nn
 
 from mmdet3d.registry import MODELS
-from .. import builder
 from .utils import VFELayer, get_paddings_indicator
 
 
@@ -172,7 +171,7 @@ class DynamicVFE(nn.Module):
             voxel_size, point_cloud_range, average_points=True)
         self.fusion_layer = None
         if fusion_layer is not None:
-            self.fusion_layer = builder.build_fusion_layer(fusion_layer)
+            self.fusion_layer = MODELS.build(fusion_layer)
 
     def map_voxel_center_to_point(self, pts_coors, voxel_mean, voxel_coors):
         """Map voxel features to its corresponding points.
@@ -381,7 +380,7 @@ class HardVFE(nn.Module):
 
         self.fusion_layer = None
         if fusion_layer is not None:
-            self.fusion_layer = builder.build_fusion_layer(fusion_layer)
+            self.fusion_layer = MODELS.build(fusion_layer)
 
     def forward(self,
                 features,

tests/test_models/test_backbones/test_dgcnn.py

@@ -3,7 +3,7 @@ import numpy as np
 import pytest
 import torch
 
-from mmdet3d.models import build_backbone
+from mmdet3d.registry import MODELS
 
 
 def test_dgcnn_gf():
@@ -21,7 +21,7 @@ def test_dgcnn_gf():
         fa_channels=(1024, ),
         act_cfg=dict(type='ReLU'))
 
-    self = build_backbone(cfg)
+    self = MODELS.build(cfg)
     self.cuda()
 
     xyz = np.fromfile('tests/data/sunrgbd/points/000001.bin', dtype=np.float32)

tests/test_models/test_backbones/test_dla.py

 import torch
 
-from mmdet3d.models import build_backbone
+from mmdet3d.registry import MODELS
 
 
 def test_dla_net():
@@ -13,7 +13,7 @@ def test_dla_net():
         norm_cfg=dict(type='GN', num_groups=32))
 
     img = torch.randn((4, 3, 32, 32))
-    self = build_backbone(cfg)
+    self = MODELS.build(cfg)
     self.init_weights()
 
     results = self(img)

tests/test_models/test_backbones/test_mink_resnet.py

@@ -3,7 +3,7 @@ import numpy as np
 import pytest
 import torch
 
-from mmdet3d.models import build_backbone
+from mmdet3d.registry import MODELS
 
 
 def test_mink_resnet():
@@ -30,7 +30,7 @@ def test_mink_resnet():
 
     # MinkResNet34 with 4 outputs
     cfg = dict(type='MinkResNet', depth=34, in_channels=3)
-    self = build_backbone(cfg).cuda()
+    self = MODELS.build(cfg).cuda()
     self.init_weights()
 
     y = self(x)
@@ -47,7 +47,7 @@ def test_mink_resnet():
     # MinkResNet50 with 2 outputs
     cfg = dict(
         type='MinkResNet', depth=34, in_channels=3, num_stages=2, pool=False)
-    self = build_backbone(cfg).cuda()
+    self = MODELS.build(cfg).cuda()
     self.init_weights()
 
     y = self(x)

tests/test_models/test_backbones/test_multi_backbone.py

@@ -3,7 +3,7 @@ import numpy as np
 import pytest
 import torch
 
-from mmdet3d.models import build_backbone
+from mmdet3d.registry import MODELS
 
 
 def test_multi_backbone():
@@ -58,7 +58,7 @@ def test_multi_backbone():
                 norm_cfg=dict(type='BN2d'))
         ])
 
-    self = build_backbone(cfg_list)
+    self = MODELS.build(cfg_list)
     self.cuda()
 
     assert len(self.backbone_list) == 4
@@ -89,7 +89,7 @@ def test_multi_backbone():
             fp_channels=((256, 256), (256, 256)),
             norm_cfg=dict(type='BN2d')))
 
-    self = build_backbone(cfg_dict)
+    self = MODELS.build(cfg_dict)
     self.cuda()
 
     assert len(self.backbone_list) == 2
@@ -104,14 +104,14 @@ def test_multi_backbone():
     # Length of backbone configs list should be equal to num_streams
     with pytest.raises(AssertionError):
         cfg_list['num_streams'] = 3
-        build_backbone(cfg_list)
+        MODELS.build(cfg_list)
 
     # Length of suffixes list should be equal to num_streams
     with pytest.raises(AssertionError):
         cfg_dict['suffixes'] = ['net0', 'net1', 'net2']
-        build_backbone(cfg_dict)
+        MODELS.build(cfg_dict)
 
     # Type of 'backbones' should be Dict or List[Dict].
     with pytest.raises(AssertionError):
         cfg_dict['backbones'] = 'PointNet2SASSG'
-        build_backbone(cfg_dict)
+        MODELS.build(cfg_dict)

tests/test_models/test_backbones/test_pointnet2_sa_msg.py

@@ -3,7 +3,7 @@ import numpy as np
 import pytest
 import torch
 
-from mmdet3d.models import build_backbone
+from mmdet3d.registry import MODELS
 
 
 def test_pointnet2_sa_msg():
@@ -30,7 +30,7 @@ def test_pointnet2_sa_msg():
             use_xyz=True,
             normalize_xyz=False))
 
-    self = build_backbone(cfg)
+    self = MODELS.build(cfg)
     self.cuda()
     assert self.SA_modules[0].mlps[0].layer0.conv.in_channels == 4
     assert self.SA_modules[0].mlps[0].layer0.conv.out_channels == 8
@@ -51,7 +51,7 @@ def test_pointnet2_sa_msg():
 
     # out_indices should smaller than the length of SA Modules.
     with pytest.raises(AssertionError):
-        build_backbone(
+        MODELS.build(
             dict(
                 type='PointNet2SAMSG',
                 in_channels=4,
@@ -95,7 +95,7 @@ def test_pointnet2_sa_msg():
             use_xyz=True,
             normalize_xyz=False))
 
-    self = build_backbone(cfg)
+    self = MODELS.build(cfg)
     self.cuda()
     ret_dict = self(xyz)
     sa_xyz = ret_dict['sa_xyz']

tests/test_models/test_backbones/test_pointnet2_sa_ssg.py

@@ -3,7 +3,7 @@ import numpy as np
 import pytest
 import torch
 
-from mmdet3d.models import build_backbone
+from mmdet3d.registry import MODELS
 
 
 def test_pointnet2_sa_ssg():
@@ -18,7 +18,7 @@ def test_pointnet2_sa_ssg():
         num_samples=(16, 8),
         sa_channels=((8, 16), (16, 16)),
         fp_channels=((16, 16), (16, 16)))
-    self = build_backbone(cfg)
+    self = MODELS.build(cfg)
     self.cuda()
     assert self.SA_modules[0].mlps[0].layer0.conv.in_channels == 6
     assert self.SA_modules[0].mlps[0].layer0.conv.out_channels == 8
@@ -61,7 +61,7 @@ def test_pointnet2_sa_ssg():
 
     # test only xyz input without features
     cfg['in_channels'] = 3
-    self = build_backbone(cfg)
+    self = MODELS.build(cfg)
     self.cuda()
     ret_dict = self(xyz[..., :3])
     assert len(fp_xyz) == len(fp_features) == len(fp_indices) == 3

tests/test_models/test_losses/test_multibin_loss.py

@@ -2,7 +2,7 @@
 import pytest
 import torch
 
-from mmdet3d.models.builder import build_loss
+from mmdet3d.registry import MODELS
 
 
 def test_multibin_loss():
@@ -24,7 +24,7 @@ def test_multibin_loss():
                            [1, 1, 0, 0, 3.12, 3.12, 2.34, 1.23]])
     multibin_loss_cfg = dict(
         type='MultiBinLoss', reduction='none', loss_weight=1.0)
-    multibin_loss = build_loss(multibin_loss_cfg)
+    multibin_loss = MODELS.build(multibin_loss_cfg)
     output_multibin_loss = multibin_loss(pred, target, num_dir_bins=4)
     expected_multibin_loss = torch.tensor(2.1120)
     assert torch.allclose(