questions #12
Description
Here are the two questions:
1、What FPS can this model achieve, and can it perform real-time inference?
2、This project and the recently open-sourced SAMURAI both seem to focus on object re-identification and tracking. I've tested SAMURAI and noticed it has instances of ID switching. Have you compared the performance of the two?