I am working on 3D object detection using tf3d.
Github repository - [login to view URL]
Want to prepare .tfrecord with my own 3D data (from Depth camera) with which i can train tf3d object detection model.
I have one sample .tfrecord file (gs://waymo_open_dataset_tf_example_tf3d/original_tfrecords/[login to view URL] .) I don't know the exact structure of this sample tfrecord file but can say it is mix of RGB and Depth images, point cloud data and camera parameters and scene information.
Exact problem can be defined as to identify the structure of sample .tfrecord ([login to view URL]) and preparing a python code to generate own .tfrecord of similar structure with my own data.