Generate the result file in JSON format
June-0814 opened this issue · 1 comments
June-0814 commented
How to extract skeleton information in batch？And how to get the result in json format？
stefanopini commented
Hi!
I've added the format option to the extract-keypoints.py script in the latest commit!
Run python scripts/extract-keypoints.py --help for more information.
Currently, the frames are processed with batch size 1 from YOLOv3, but then detected people are processed in batch (i.e. the batch size of the input of HRNet is the number of detected people in each frame).
To change this, you could load multiple frames, stack them together and call model.predict(frames) on the stacked frames.