search

bowenc0221 / panoptic-deeplab

This is Pytorch re-implementation of our CVPR 2020 paper "Panoptic-DeepLab: A Simple, Strong, and Fast Baseline for Bottom-Up Panoptic Segmentation" (https://arxiv.org/abs/1911.10194)
Apache License 2.0
585 stars 117 forks source link

How to generate json files? #66

Closed lxa9867 closed 3 years ago

lxa9867 commented 3 years ago

Hello, thank you for sharing this great work.

I am trying to reproduce some results on Cityscapes dataset. And I didn't find any files that can generate json file in COCO format. Does the current codebase support to generate json file for submitting? Thank you.

bowenc0221 commented 3 years ago

Do you mean the training annotation json files? They are generated by CITYSCAPES_DATASET=/path/to/abovementioned/cityscapes python cityscapesscripts/preparation/createPanopticImgs.py, please read the README for more detail.

lxa9867 commented 3 years ago

Do you mean the training annotation json files? They are generated by CITYSCAPES_DATASET=/path/to/abovementioned/cityscapes python cityscapesscripts/preparation/createPanopticImgs.py, please read the README for more detail.

Thank you for your reply. While I mean how to generate json files from the prediction results? I am trying to reproduce results with MV and pseudo labels (naive-student) on Cityscapes. I hope to iteratively train the model so I need to prepare the output predicts have the same format as COCO format annotations. Is there any sources can be referred? Thank you!

bowenc0221 commented 3 years ago

This file (https://github.com/bowenc0221/panoptic-deeplab/blob/master/segmentation/evaluation/panoptic.py) writes prediction to json for evaluation.