Only used for self-learning
I implement the method Seq-NMS mentioned in paper: SEQ-NMS FOR VIDEO OBJECT DETECTION based on the Faster-RCNN.
As for the detection in a single frame, I use the demo program and pre-trained network given by Faster-RCNN to run the detection in each frame in a video.
Follow the demo instruction given by Faster-RCNN, and replace the demo.py file by the seqnms.py given here.