visulization

[yujialele]

Hi! Thank you for your good work! I obtain the prediction result using the validation codes and find that each segments in the video are shown. Therefore, I want to know how to generate the final each event instance boundaries. Do you need to combine multi-segment predictions under the same label ?