Closed dyther closed 5 years ago
You need to retrain the model on another video. If you want to use the trained model on any video, it requires to train this model on a very large number of videos.
So,it cant be used for real-time detection? for the LK Operationr is a little slow
After training, the learned detector can be used for real-time detection, since the inference procedure does not require the LK operator. (We are working on improving this algorithm including speeding up the LK, and will be public available soon)
Thanks for your great project. I have run the demo_sbr.sh and I can see the cpm detector has improved a lot on demo-sbr,mp4. Now I want to use another video to test the detector , can I directly use the existent model? or should I retrain the model on another video? If I can use the trained model on any videos?