Closed haoxurt closed 2 years ago
@haoxurt
third_view
assumes arms are visible and we will use arm information to help hand detectors (majorly use arm information to distinguish left/right hands, which are often wrong predicted by hand detectors). ego_centric
assumes no arms visible, therefore, only rely on hand detectors' predictions.
what's the difference of ego_centric and third_view hand detector? Which should I choose when I test with random image?