DeepLab2 is a TensorFlow library for deep labeling, aiming to provide a unified and state-of-the-art TensorFlow codebase for dense pixel labeling tasks.
Hi, you mentioned in 'data/dataset.py' that the model supports Panoramic Video Panoptic Segmentation (PVPS) of the Waymo Open Dataset (WOD). I'm wondering how this is possible since I don't see any executable command or example for training on *.parquet files (provided by version 2.0 of WOD). Should I extract the images+annotations and then convert them into TFRecords?
Could you please advise me on how to train DeepLab2 on WOD using PVPS?
Hi, you mentioned in 'data/dataset.py' that the model supports Panoramic Video Panoptic Segmentation (PVPS) of the Waymo Open Dataset (WOD). I'm wondering how this is possible since I don't see any executable command or example for training on *.parquet files (provided by version 2.0 of WOD). Should I extract the images+annotations and then convert them into TFRecords?
Could you please advise me on how to train DeepLab2 on WOD using PVPS?