Today Waymo open its dataset with a script to view the data. I followed the tutorials on their github webpage and found that I can't build the bazel project on my machine. However, I need some python scripts in order to view the dataset, so I use their colab project to build the python script. If you have the same issue like "ImportError: cannot import name 'dataset_pb2'" or "ImportError: cannot import name 'label_pb2'", download the two scripts here.
Code in test.py are mainly from their tutorial notebook, I add some salt to make it work with the dataset you get from waymo website.
1. Download Waymo's github code [here](https://github.com/waymo-research/waymo-open-dataset)
2. Put dataset_pb2.py and label_pb2.py in waymo_open_dataset folder
3. Change python path and waymo dataset path in test.py
4. Run test.py
Here are some projects I found useful while processing the Waymo dataset. Please check.
1. Waymo Open Dataset Viewer
[https://github.com/erksch/waymo-open-dataset-viewer](https://github.com/erksch/waymo-open-dataset-viewer)
A WebGL viewer for pointclouds of the waymo open dataset that runs seamlessly in the browser with an integrated python server that processes and serves the dataset segments.
2. Waymo_Kitti_Adapter
[https://github.com/Yao-Shao/Waymo_Kitti_Adapter](https://github.com/Yao-Shao/Waymo_Kitti_Adapter)
This is a tool converting Waymo open dataset format to Kitti dataset format.
3. Simple Waymo Open Dataset Reader
[https://github.com/gdlg/simple-waymo-open-dataset-reader](https://github.com/gdlg/simple-waymo-open-dataset-reader)
This is a simple file reader for the Waymo Open Dataset which does not depend on TensorFlow and Bazel. The main goal is to be able to quickly integrate Waymo’s dataset with other deep learning frameworks without having to pull tons of dependencies. It does not aim to replace the whole framework, especially the evaluation metrics that they provide.