[Enhance] Support load annotations from ceph (#1325)

* support load annotations from ceph in mmdet3d

* update doc about ceph backend

* fix comments

docs/en/tutorials/backends_support.md

+# Tutorial 7: Backends Support
+
+We support different file client backends: Disk, Ceph and LMDB, etc. Here is an example of how to modify configs for Ceph-based data loading and saving.
+
+## Load data and annotations from Ceph
+
+We support loading data and generated annotation info files (pkl and json) from Ceph:
+
+```python
+# set file client backends as Ceph
+file_client_args = dict(
+    backend='petrel',
+    path_mapping=dict({
+        './data/nuscenes/':
+        's3://openmmlab/datasets/detection3d/nuscenes/', # replace the path with your data path on Ceph
+        'data/nuscenes/':
+        's3://openmmlab/datasets/detection3d/nuscenes/' # replace the path with your data path on Ceph
+    }))
+
+db_sampler = dict(
+    data_root=data_root,
+    info_path=data_root + 'kitti_dbinfos_train.pkl',
+    rate=1.0,
+    prepare=dict(filter_by_difficulty=[-1], filter_by_min_points=dict(Car=5)),
+    sample_groups=dict(Car=15),
+    classes=class_names,
+    # set file client for points loader to load training data
+    points_loader=dict(
+        type='LoadPointsFromFile',
+        coord_type='LIDAR',
+        load_dim=4,
+        use_dim=4,
+        file_client_args=file_client_args),
+    # set file client for data base sampler to load db info file
+    file_client_args=file_client_args)
+
+train_pipeline = [
+    # set file client for loading training data
+    dict(type='LoadPointsFromFile', coord_type='LIDAR', load_dim=4, use_dim=4, file_client_args=file_client_args),
+    # set file client for loading training data annotations
+    dict(type='LoadAnnotations3D', with_bbox_3d=True, with_label_3d=True, file_client_args=file_client_args),
+    dict(type='ObjectSample', db_sampler=db_sampler),
+    dict(
+        type='ObjectNoise',
+        num_try=100,
+        translation_std=[0.25, 0.25, 0.25],
+        global_rot_range=[0.0, 0.0],
+        rot_range=[-0.15707963267, 0.15707963267]),
+    dict(type='RandomFlip3D', flip_ratio_bev_horizontal=0.5),
+    dict(
+        type='GlobalRotScaleTrans',
+        rot_range=[-0.78539816, 0.78539816],
+        scale_ratio_range=[0.95, 1.05]),
+    dict(type='PointsRangeFilter', point_cloud_range=point_cloud_range),
+    dict(type='ObjectRangeFilter', point_cloud_range=point_cloud_range),
+    dict(type='PointShuffle'),
+    dict(type='DefaultFormatBundle3D', class_names=class_names),
+    dict(type='Collect3D', keys=['points', 'gt_bboxes_3d', 'gt_labels_3d'])
+]
+test_pipeline = [
+    # set file client for loading validation/testing data
+    dict(type='LoadPointsFromFile', coord_type='LIDAR', load_dim=4, use_dim=4, file_client_args=file_client_args),
+    dict(
+        type='MultiScaleFlipAug3D',
+        img_scale=(1333, 800),
+        pts_scale_ratio=1,
+        flip=False,
+        transforms=[
+            dict(
+                type='GlobalRotScaleTrans',
+                rot_range=[0, 0],
+                scale_ratio_range=[1., 1.],
+                translation_std=[0, 0, 0]),
+            dict(type='RandomFlip3D'),
+            dict(
+                type='PointsRangeFilter', point_cloud_range=point_cloud_range),
+            dict(
+                type='DefaultFormatBundle3D',
+                class_names=class_names,
+                with_label=False),
+            dict(type='Collect3D', keys=['points'])
+        ])
+]
+
+data = dict(
+    # set file client for loading training info files (.pkl)
+    train=dict(
+        type='RepeatDataset',
+        times=2,
+        dataset=dict(pipeline=train_pipeline, classes=class_names, file_client_args=file_client_args)),
+    # set file client for loading validation info files (.pkl)
+    val=dict(pipeline=test_pipeline, classes=class_names,file_client_args=file_client_args),
+    # set file client for loading testing info files (.pkl)
+    test=dict(pipeline=test_pipeline, classes=class_names, file_client_args=file_client_args))
+```
+
+
+
+## Load pretrained model from Ceph
+```python
+model = dict(
+    pts_backbone=dict(
+        _delete_=True,
+        type='NoStemRegNet',
+        arch='regnetx_1.6gf',
+        init_cfg=dict(
+            type='Pretrained', checkpoint='s3://openmmlab/checkpoints/mmdetection3d/regnetx_1.6gf'), # replace the path with your pretrained model path on Ceph
+        ...
+```
+
+## Load checkpoint from Ceph
+```python
+# replace the path with your checkpoint path on Ceph
+load_from = 's3://openmmlab/checkpoints/mmdetection3d/v0.1.0_models/pointpillars/hv_pointpillars_secfpn_6x8_160e_kitti-3d-car/hv_pointpillars_secfpn_6x8_160e_kitti-3d-car_20200620_230614-77663cd6.pth.pth'
+resume_from = None
+workflow = [('train', 1)]
+```
+
+## Save checkpoint into Ceph
+
+```python
+# checkpoint saving
+# replace the path with your checkpoint saving path on Ceph
+checkpoint_config = dict(interval=1, max_keep_ckpts=2, out_dir='s3://openmmlab/mmdetection3d')
+```
+
+## EvalHook saves the best checkpoint into Ceph
+
+```python
+# replace the path with your checkpoint saving path on Ceph
+evaluation = dict(interval=1, save_best='bbox', out_dir='s3://openmmlab/mmdetection3d')
+```
+
+## Save the training log into Ceph
+The training log will be backed up to the specified Ceph path after training.
+
+```python
+log_config = dict(
+    interval=50,
+    hooks=[
+        dict(type='TextLoggerHook', out_dir='s3://openmmlab/mmdetection3d'),
+    ])
+```
+You can also delete the local training log after backing up to the specified Ceph path by setting `keep_local = False`.
+
+```python
+log_config = dict(
+    interval=50,
+    hooks=[
+        dict(type='TextLoggerHook', out_dir='s3://openmmlab/mmdetection3d'', keep_local=False),
+    ])
+```

docs/en/tutorials/index.rst

@@ -7,3 +7,4 @@
    customize_models.md
    customize_runtime.md
    coord_sys_tutorial.md
+   backends_support.md

docs/zh_cn/tutorials/backends_support.md

+# Tutorial 7: 后端支持
+我们支持不同的文件客户端后端：磁盘、Ceph 和 LMDB 等。下面是修改配置使之从 Ceph 加载和保存数据的示例。
+
+## 从 Ceph 读取数据和标注文件
+
+我们支持从 Ceph 加载数据和生成的标注信息文件（pkl 和 json）：
+
+```python
+# set file client backends as Ceph
+file_client_args = dict(
+    backend='petrel',
+    path_mapping=dict({
+        './data/nuscenes/':
+        's3://openmmlab/datasets/detection3d/nuscenes/', # replace the path with your data path on Ceph
+        'data/nuscenes/':
+        's3://openmmlab/datasets/detection3d/nuscenes/' # replace the path with your data path on Ceph
+    }))
+
+db_sampler = dict(
+    data_root=data_root,
+    info_path=data_root + 'kitti_dbinfos_train.pkl',
+    rate=1.0,
+    prepare=dict(filter_by_difficulty=[-1], filter_by_min_points=dict(Car=5)),
+    sample_groups=dict(Car=15),
+    classes=class_names,
+    # set file client for points loader to load training data
+    points_loader=dict(
+        type='LoadPointsFromFile',
+        coord_type='LIDAR',
+        load_dim=4,
+        use_dim=4,
+        file_client_args=file_client_args),
+    # set file client for data base sampler to load db info file
+    file_client_args=file_client_args)
+
+train_pipeline = [
+    # set file client for loading training data
+    dict(type='LoadPointsFromFile', coord_type='LIDAR', load_dim=4, use_dim=4, file_client_args=file_client_args),
+    # set file client for loading training data annotations
+    dict(type='LoadAnnotations3D', with_bbox_3d=True, with_label_3d=True, file_client_args=file_client_args),
+    dict(type='ObjectSample', db_sampler=db_sampler),
+    dict(
+        type='ObjectNoise',
+        num_try=100,
+        translation_std=[0.25, 0.25, 0.25],
+        global_rot_range=[0.0, 0.0],
+        rot_range=[-0.15707963267, 0.15707963267]),
+    dict(type='RandomFlip3D', flip_ratio_bev_horizontal=0.5),
+    dict(
+        type='GlobalRotScaleTrans',
+        rot_range=[-0.78539816, 0.78539816],
+        scale_ratio_range=[0.95, 1.05]),
+    dict(type='PointsRangeFilter', point_cloud_range=point_cloud_range),
+    dict(type='ObjectRangeFilter', point_cloud_range=point_cloud_range),
+    dict(type='PointShuffle'),
+    dict(type='DefaultFormatBundle3D', class_names=class_names),
+    dict(type='Collect3D', keys=['points', 'gt_bboxes_3d', 'gt_labels_3d'])
+]
+test_pipeline = [
+    # set file client for loading validation/testing data
+    dict(type='LoadPointsFromFile', coord_type='LIDAR', load_dim=4, use_dim=4, file_client_args=file_client_args),
+    dict(
+        type='MultiScaleFlipAug3D',
+        img_scale=(1333, 800),
+        pts_scale_ratio=1,
+        flip=False,
+        transforms=[
+            dict(
+                type='GlobalRotScaleTrans',
+                rot_range=[0, 0],
+                scale_ratio_range=[1., 1.],
+                translation_std=[0, 0, 0]),
+            dict(type='RandomFlip3D'),
+            dict(
+                type='PointsRangeFilter', point_cloud_range=point_cloud_range),
+            dict(
+                type='DefaultFormatBundle3D',
+                class_names=class_names,
+                with_label=False),
+            dict(type='Collect3D', keys=['points'])
+        ])
+]
+
+data = dict(
+    # set file client for loading training info files (.pkl)
+    train=dict(
+        type='RepeatDataset',
+        times=2,
+        dataset=dict(pipeline=train_pipeline, classes=class_names, file_client_args=file_client_args)),
+    # set file client for loading validation info files (.pkl)
+    val=dict(pipeline=test_pipeline, classes=class_names,file_client_args=file_client_args),
+    # set file client for loading testing info files (.pkl)
+    test=dict(pipeline=test_pipeline, classes=class_names, file_client_args=file_client_args))
+```
+
+
+
+## 从 Ceph 读取预训练模型
+```python
+model = dict(
+    pts_backbone=dict(
+        _delete_=True,
+        type='NoStemRegNet',
+        arch='regnetx_1.6gf',
+        init_cfg=dict(
+            type='Pretrained', checkpoint='s3://openmmlab/checkpoints/mmdetection3d/regnetx_1.6gf'), # replace the path with your pretrained model path on Ceph
+        ...
+```
+
+## 从 Ceph 读取模型权重文件
+```python
+# replace the path with your checkpoint path on Ceph
+load_from = 's3://openmmlab/checkpoints/mmdetection3d/v0.1.0_models/pointpillars/hv_pointpillars_secfpn_6x8_160e_kitti-3d-car/hv_pointpillars_secfpn_6x8_160e_kitti-3d-car_20200620_230614-77663cd6.pth.pth'
+resume_from = None
+workflow = [('train', 1)]
+```
+
+## 保存模型权重文件至 Ceph
+
+```python
+# checkpoint saving
+# replace the path with your checkpoint saving path on Ceph
+checkpoint_config = dict(interval=1, max_keep_ckpts=2, out_dir='s3://openmmlab/mmdetection3d')
+```
+
+## EvalHook 保存最优模型权重文件至 Ceph
+
+```python
+# replace the path with your checkpoint saving path on Ceph
+evaluation = dict(interval=1, save_best='bbox', out_dir='s3://openmmlab/mmdetection3d')
+```
+
+## 训练日志保存至 Ceph
+训练后的训练日志会备份到指定的 Ceph 路径。
+
+```python
+log_config = dict(
+    interval=50,
+    hooks=[
+        dict(type='TextLoggerHook', out_dir='s3://openmmlab/mmdetection3d'),
+    ])
+```
+您还可以通过设置 `keep_local = False` 备份到指定的 Ceph 路径后删除本地训练日志。
+
+```python
+log_config = dict(
+    interval=50,
+    hooks=[
+        dict(type='TextLoggerHook', out_dir='s3://openmmlab/mmdetection3d'', keep_local=False),
+    ])
+```

docs/zh_cn/tutorials/index.rst

@@ -7,3 +7,4 @@
    customize_models.md
    customize_runtime.md
    coord_sys_tutorial.md
+   backends_support.md

mmdet3d/datasets/custom_3d.py

@@ -51,7 +51,8 @@ class Custom3DDataset(Dataset):
                  modality=None,
                  box_type_3d='LiDAR',
                  filter_empty_gt=True,
-                 test_mode=False):
+                 test_mode=False,
+                 file_client_args=dict(backend='disk')):
         super().__init__()
         self.data_root = data_root
         self.ann_file = ann_file
@@ -61,13 +62,26 @@ class Custom3DDataset(Dataset):
         self.box_type_3d, self.box_mode_3d = get_box_type(box_type_3d)
 
         self.CLASSES = self.get_classes(classes)
+        self.file_client = mmcv.FileClient(**file_client_args)
         self.cat2id = {name: i for i, name in enumerate(self.CLASSES)}
-        self.data_infos = self.load_annotations(self.ann_file)
 
+        # load annotations
+        if hasattr(self.file_client, 'get_local_path'):
+            with self.file_client.get_local_path(self.ann_file) as local_path:
+                self.data_infos = self.load_annotations(open(local_path, 'rb'))
+        else:
+            warnings.warn(
+                'The used MMCV version does not have get_local_path. '
+                f'We treat the {self.ann_file} as local paths and it '
+                'might cause errors if the path is not a local path. '
+                'Please use MMCV>= 1.3.16 if you meet errors.')
+            self.data_infos = self.load_annotations(self.ann_file)
+
+        # process pipeline
         if pipeline is not None:
             self.pipeline = Compose(pipeline)
 
-        # set group flag for the sampler
+        # set group flag for the samplers
         if not self.test_mode:
             self._set_group_flag()
 
@@ -80,7 +94,8 @@ class Custom3DDataset(Dataset):
         Returns:
             list[dict]: List of annotations.
         """
-        return mmcv.load(ann_file)
+        # loading data from a file-like object needs file format
+        return mmcv.load(ann_file, file_format='pkl')
 
     def get_data_info(self, index):
         """Get data info according to the given index.

mmdet3d/datasets/custom_3d_seg.py

@@ -62,14 +62,26 @@ class Custom3DSegDataset(Dataset):
                  modality=None,
                  test_mode=False,
                  ignore_index=None,
-                 scene_idxs=None):
+                 scene_idxs=None,
+                 file_client_args=dict(backend='disk')):
         super().__init__()
         self.data_root = data_root
         self.ann_file = ann_file
         self.test_mode = test_mode
         self.modality = modality
+        self.file_client = mmcv.FileClient(**file_client_args)
 
-        self.data_infos = self.load_annotations(self.ann_file)
+        # load annotations
+        if hasattr(self.file_client, 'get_local_path'):
+            with self.file_client.get_local_path(self.ann_file) as local_path:
+                self.data_infos = self.load_annotations(open(local_path, 'rb'))
+        else:
+            warnings.warn(
+                'The used MMCV version does not have get_local_path. '
+                f'We treat the {self.ann_file} as local paths and it '
+                'might cause errors if the path is not a local path. '
+                'Please use MMCV>= 1.3.16 if you meet errors.')
+            self.data_infos = self.load_annotations(self.ann_file)
 
         if pipeline is not None:
             self.pipeline = Compose(pipeline)
@@ -94,7 +106,8 @@ class Custom3DSegDataset(Dataset):
         Returns:
             list[dict]: List of annotations.
         """
-        return mmcv.load(ann_file)
+        # loading data from a file-like object needs file format
+        return mmcv.load(ann_file, file_format='pkl')
 
     def get_data_info(self, index):
         """Get data info according to the given index.

mmdet3d/datasets/kitti_dataset.py

@@ -65,7 +65,8 @@ class KittiDataset(Custom3DDataset):
                  box_type_3d='LiDAR',
                  filter_empty_gt=True,
                  test_mode=False,
-                 pcd_limit_range=[0, -40, -3, 70.4, 40, 0.0]):
+                 pcd_limit_range=[0, -40, -3, 70.4, 40, 0.0],
+                 **kwargs):
         super().__init__(
             data_root=data_root,
             ann_file=ann_file,
@@ -74,7 +75,8 @@ class KittiDataset(Custom3DDataset):
             modality=modality,
             box_type_3d=box_type_3d,
             filter_empty_gt=filter_empty_gt,
-            test_mode=test_mode)
+            test_mode=test_mode,
+            **kwargs)
 
         self.split = split
         self.root_split = os.path.join(self.data_root, split)

mmdet3d/datasets/lyft_dataset.py

@@ -86,7 +86,8 @@ class LyftDataset(Custom3DDataset):
                  modality=None,
                  box_type_3d='LiDAR',
                  filter_empty_gt=True,
-                 test_mode=False):
+                 test_mode=False,
+                 **kwargs):
         self.load_interval = load_interval
         super().__init__(
             data_root=data_root,
@@ -96,7 +97,8 @@ class LyftDataset(Custom3DDataset):
             modality=modality,
             box_type_3d=box_type_3d,
             filter_empty_gt=filter_empty_gt,
-            test_mode=test_mode)
+            test_mode=test_mode,
+            **kwargs)
 
         if self.modality is None:
             self.modality = dict(
@@ -116,7 +118,8 @@ class LyftDataset(Custom3DDataset):
         Returns:
             list[dict]: List of annotations sorted by timestamps.
         """
-        data = mmcv.load(ann_file)
+        # loading data from a file-like object needs file format
+        data = mmcv.load(ann_file, file_format='pkl')
         data_infos = list(sorted(data['infos'], key=lambda e: e['timestamp']))
         data_infos = data_infos[::self.load_interval]
         self.metadata = data['metadata']

mmdet3d/datasets/nuscenes_dataset.py

@@ -125,7 +125,8 @@ class NuScenesDataset(Custom3DDataset):
                  filter_empty_gt=True,
                  test_mode=False,
                  eval_version='detection_cvpr_2019',
-                 use_valid_flag=False):
+                 use_valid_flag=False,
+                 **kwargs):
         self.load_interval = load_interval
         self.use_valid_flag = use_valid_flag
         super().__init__(
@@ -136,7 +137,8 @@ class NuScenesDataset(Custom3DDataset):
             modality=modality,
             box_type_3d=box_type_3d,
             filter_empty_gt=filter_empty_gt,
-            test_mode=test_mode)
+            test_mode=test_mode,
+            **kwargs)
 
         self.with_velocity = with_velocity
         self.eval_version = eval_version
@@ -184,7 +186,8 @@ class NuScenesDataset(Custom3DDataset):
         Returns:
             list[dict]: List of annotations sorted by timestamps.
         """
-        data = mmcv.load(ann_file)
+        # loading data from a file-like object needs file format
+        data = mmcv.load(ann_file, file_format='pkl')
         data_infos = list(sorted(data['infos'], key=lambda e: e['timestamp']))
         data_infos = data_infos[::self.load_interval]
         self.metadata = data['metadata']

mmdet3d/datasets/pipelines/dbsampler.py

 # Copyright (c) OpenMMLab. All rights reserved.
 import copy
 import os
+import warnings
 
 import mmcv
 import numpy as np
@@ -104,7 +105,8 @@ class DataBaseSampler(object):
                      type='LoadPointsFromFile',
                      coord_type='LIDAR',
                      load_dim=4,
-                     use_dim=[0, 1, 2, 3])):
+                     use_dim=[0, 1, 2, 3]),
+                 file_client_args=dict(backend='disk')):
         super().__init__()
         self.data_root = data_root
         self.info_path = info_path
@@ -114,8 +116,20 @@ class DataBaseSampler(object):
         self.cat2label = {name: i for i, name in enumerate(classes)}
         self.label2cat = {i: name for i, name in enumerate(classes)}
         self.points_loader = mmcv.build_from_cfg(points_loader, PIPELINES)
+        self.file_client = mmcv.FileClient(**file_client_args)
 
-        db_infos = mmcv.load(info_path)
+        # load data base infos
+        if hasattr(self.file_client, 'get_local_path'):
+            with self.file_client.get_local_path(info_path) as local_path:
+                # loading data from a file-like object needs file format
+                db_infos = mmcv.load(open(local_path, 'rb'), file_format='pkl')
+        else:
+            warnings.warn(
+                'The used MMCV version does not have get_local_path. '
+                f'We treat the {info_path} as local paths and it '
+                'might cause errors if the path is not a local path. '
+                'Please use MMCV>= 1.3.16 if you meet errors.')
+            db_infos = mmcv.load(info_path)
 
         # filter database infos
         from mmdet3d.utils import get_root_logger

mmdet3d/datasets/s3dis_dataset.py

@@ -54,7 +54,8 @@ class S3DISDataset(Custom3DDataset):
                  modality=None,
                  box_type_3d='Depth',
                  filter_empty_gt=True,
-                 test_mode=False):
+                 test_mode=False,
+                 *kwargs):
         super().__init__(
             data_root=data_root,
             ann_file=ann_file,
@@ -63,7 +64,8 @@ class S3DISDataset(Custom3DDataset):
             modality=modality,
             box_type_3d=box_type_3d,
             filter_empty_gt=filter_empty_gt,
-            test_mode=test_mode)
+            test_mode=test_mode,
+            *kwargs)
 
     def get_ann_info(self, index):
         """Get annotation info according to the given index.
@@ -205,7 +207,8 @@ class _S3DISSegDataset(Custom3DSegDataset):
                  modality=None,
                  test_mode=False,
                  ignore_index=None,
-                 scene_idxs=None):
+                 scene_idxs=None,
+                 **kwargs):
 
         super().__init__(
             data_root=data_root,
@@ -216,7 +219,8 @@ class _S3DISSegDataset(Custom3DSegDataset):
             modality=modality,
             test_mode=test_mode,
             ignore_index=ignore_index,
-            scene_idxs=scene_idxs)
+            scene_idxs=scene_idxs,
+            **kwargs)
 
     def get_ann_info(self, index):
         """Get annotation info according to the given index.
@@ -347,7 +351,8 @@ class S3DISSegDataset(_S3DISSegDataset):
                  modality=None,
                  test_mode=False,
                  ignore_index=None,
-                 scene_idxs=None):
+                 scene_idxs=None,
+                 **kwargs):
 
         # make sure that ann_files and scene_idxs have same length
         ann_files = self._check_ann_files(ann_files)
@@ -363,7 +368,8 @@ class S3DISSegDataset(_S3DISSegDataset):
             modality=modality,
             test_mode=test_mode,
             ignore_index=ignore_index,
-            scene_idxs=scene_idxs[0])
+            scene_idxs=scene_idxs[0],
+            **kwargs)
 
         datasets = [
             _S3DISSegDataset(

mmdet3d/datasets/scannet_dataset.py

@@ -58,7 +58,8 @@ class ScanNetDataset(Custom3DDataset):
                  modality=dict(use_camera=False, use_depth=True),
                  box_type_3d='Depth',
                  filter_empty_gt=True,
-                 test_mode=False):
+                 test_mode=False,
+                 **kwargs):
         super().__init__(
             data_root=data_root,
             ann_file=ann_file,
@@ -67,7 +68,8 @@ class ScanNetDataset(Custom3DDataset):
             modality=modality,
             box_type_3d=box_type_3d,
             filter_empty_gt=filter_empty_gt,
-            test_mode=test_mode)
+            test_mode=test_mode,
+            **kwargs)
         assert 'use_camera' in self.modality and \
                'use_depth' in self.modality
         assert self.modality['use_camera'] or self.modality['use_depth']
@@ -322,7 +324,8 @@ class ScanNetSegDataset(Custom3DSegDataset):
                  modality=None,
                  test_mode=False,
                  ignore_index=None,
-                 scene_idxs=None):
+                 scene_idxs=None,
+                 **kwargs):
 
         super().__init__(
             data_root=data_root,
@@ -333,7 +336,8 @@ class ScanNetSegDataset(Custom3DSegDataset):
             modality=modality,
             test_mode=test_mode,
             ignore_index=ignore_index,
-            scene_idxs=scene_idxs)
+            scene_idxs=scene_idxs,
+            **kwargs)
 
     def get_ann_info(self, index):
         """Get annotation info according to the given index.

mmdet3d/datasets/sunrgbd_dataset.py

@@ -54,7 +54,8 @@ class SUNRGBDDataset(Custom3DDataset):
                  modality=dict(use_camera=True, use_lidar=True),
                  box_type_3d='Depth',
                  filter_empty_gt=True,
-                 test_mode=False):
+                 test_mode=False,
+                 **kwargs):
         super().__init__(
             data_root=data_root,
             ann_file=ann_file,
@@ -63,7 +64,8 @@ class SUNRGBDDataset(Custom3DDataset):
             modality=modality,
             box_type_3d=box_type_3d,
             filter_empty_gt=filter_empty_gt,
-            test_mode=test_mode)
+            test_mode=test_mode,
+            **kwargs)
         assert 'use_camera' in self.modality and \
             'use_lidar' in self.modality
         assert self.modality['use_camera'] or self.modality['use_lidar']

mmdet3d/datasets/waymo_dataset.py

@@ -66,7 +66,8 @@ class WaymoDataset(KittiDataset):
                  filter_empty_gt=True,
                  test_mode=False,
                  load_interval=1,
-                 pcd_limit_range=[-85, -85, -5, 85, 85, 5]):
+                 pcd_limit_range=[-85, -85, -5, 85, 85, 5],
+                 **kwargs):
         super().__init__(
             data_root=data_root,
             ann_file=ann_file,
@@ -78,7 +79,8 @@ class WaymoDataset(KittiDataset):
             box_type_3d=box_type_3d,
             filter_empty_gt=filter_empty_gt,
             test_mode=test_mode,
-            pcd_limit_range=pcd_limit_range)
+            pcd_limit_range=pcd_limit_range,
+            **kwargs)
 
         # to load a subset, just set the load_interval in the dataset config
         self.data_infos = self.data_infos[::load_interval]