Open lucasjinreal opened 4 years ago
Hi, @jinfagang. To use this package, you need to first do some preprocessing on the input frames, and yes, the person boxes are detected on the center frame by a detector. We will add more detailed documents to illustrate this later. Currently, you could simply try the demo program in the standalone version of AlphAction. Hope this reply solves your problem.
why not speak chinese?
Is there any demo to inference action recognition on a local video file?
How get these person boxes from videos, by a detector?