[Enhance] Reduce requirements if waymo dataset is not used (#121)

* Reduce requirements if waymo dataset is not used

* Update waymo.md

* Update optional.txt

* Update waymo.md

Add instructions of requirements/optional.txt related to waymo

docs/tutorials/waymo.md

@@ -4,6 +4,23 @@ This page provides specific tutorials about the usage of MMDetection3D for waymo
 
 ## Prepare datasets
 
+Before preparing waymo dataset, if you only installed requirements in `requirements/build.txt` and `requirements/runtime.txt` before, please install the official package for this dataset at first by running
+
+```
+# tf 2.1.0.
+pip install waymo-open-dataset-tf-2-1-0==1.2.0
+# tf 2.0.0
+# pip install waymo-open-dataset-tf-2-0-0==1.2.0
+# tf 1.15.0
+# pip install waymo-open-dataset-tf-1-15-0==1.2.0
+```
+
+or
+
+```
+pip install -r requirements/optional.txt
+```
+
 Like the general way to prepare dataset, it is recommended to symlink the dataset root to `$MMDETECTION3D/data`.
 Due to the original waymo data format is based on `tfrecord`, we need to preprocess the raw data for convenient usage in the training and evaluation procedure. Our approach is to convert them into KITTI format.
 

mmdet3d/core/evaluation/waymo_utils/prediction_kitti_to_waymo.py

@@ -2,12 +2,18 @@ r"""Adapted from `Waymo to KITTI converter
     <https://github.com/caizhongang/waymo_kitti_converter>`_.
 """
 
+try:
+    from waymo_open_dataset import dataset_pb2 as open_dataset
+except ImportError:
+    raise ImportError(
+        'Please run "pip install waymo-open-dataset-tf-2-1-0==1.2.0" '
+        'to install the official devkit first.')
+
 import mmcv
 import numpy as np
 import tensorflow as tf
 from glob import glob
 from os.path import join
-from waymo_open_dataset import dataset_pb2 as open_dataset
 from waymo_open_dataset import label_pb2
 from waymo_open_dataset.protos import metrics_pb2
 
@@ -100,8 +106,8 @@ class KITTI2Waymo(object):
         """
 
         def parse_one_object(instance_idx):
-            """Parse one instance in kitti format and convert them to
-            `Object` proto.
+            """Parse one instance in kitti format and convert them to `Object`
+            proto.
 
             Args:
                 instance_idx (int): Index of the instance to be converted.

mmdet3d/datasets/waymo_dataset.py

@@ -8,7 +8,6 @@ from os import path as osp
 
 from mmdet.datasets import DATASETS
 from ..core.bbox import Box3DMode, points_cam2img
-from ..core.evaluation.waymo_utils.prediction_kitti_to_waymo import KITTI2Waymo
 from .kitti_dataset import KittiDataset
 
 
@@ -184,6 +183,8 @@ class WaymoDataset(KittiDataset):
                                                   pklfile_prefix,
                                                   submission_prefix)
         if 'waymo' in data_format:
+            from ..core.evaluation.waymo_utils.prediction_kitti_to_waymo import \
+                KITTI2Waymo  # noqa
             waymo_root = osp.join(
                 self.data_root.split('kitti_format')[0], 'waymo_format')
             if self.split == 'training':

requirements/optional.txt

+waymo-open-dataset-tf-2-1-0==1.2.0

requirements/runtime.txt

@@ -4,8 +4,7 @@ networkx>=2.2,<2.3
 numba==0.48.0
 nuscenes-devkit
 plyfile
+scikit-image
 # by default we also use tensorboard to log results
 tensorboard
 trimesh>=2.35.39,<2.35.40
-scikit-image
-waymo-open-dataset-tf-2-1-0==1.2.0

tools/create_data.py

@@ -5,7 +5,6 @@ from tools.data_converter import indoor_converter as indoor
 from tools.data_converter import kitti_converter as kitti
 from tools.data_converter import lyft_converter as lyft_converter
 from tools.data_converter import nuscenes_converter as nuscenes_converter
-from tools.data_converter import waymo_converter as waymo
 from tools.data_converter.create_gt_database import create_groundtruth_database
 
 
@@ -150,6 +149,8 @@ def waymo_data_prep(root_path,
         max_sweeps (int): Number of input consecutive frames. Default: 5 \
             Here we store pose information of these frames for later use.
     """
+    from tools.data_converter import waymo_converter as waymo
+
     splits = ['training', 'validation', 'testing']
     for i, split in enumerate(splits):
         load_dir = osp.join(root_path, 'waymo_format', split)

tools/data_converter/waymo_converter.py

@@ -2,13 +2,18 @@ r"""Adapted from `Waymo to KITTI converter
     <https://github.com/caizhongang/waymo_kitti_converter>`_.
 """
 
+try:
+    from waymo_open_dataset import dataset_pb2
+except ImportError:
+    raise ImportError(
+        'Please run "pip install waymo-open-dataset-tf-2-1-0==1.2.0" '
+        'to install the official devkit first.')
+
 import mmcv
 import numpy as np
 import tensorflow as tf
 from glob import glob
 from os.path import join
-from waymo_open_dataset import dataset_pb2
-from waymo_open_dataset import dataset_pb2 as open_dataset
 from waymo_open_dataset.utils import range_image_utils, transform_utils
 from waymo_open_dataset.utils.frame_utils import \
     parse_range_image_and_camera_projection
@@ -101,7 +106,7 @@ class Waymo2KITTI(object):
 
         for frame_idx, data in enumerate(dataset):
 
-            frame = open_dataset.Frame()
+            frame = dataset_pb2.Frame()
             frame.ParseFromString(bytearray(data.numpy()))
             if (self.selected_waymo_locations is not None
                     and frame.context.stats.location