Video Sequence Segmentation

Step-1)
After divide the video into image frames. Use deep learning models that perform object detection, place recognition, and behavioral classification detect and store these elements in word form per frame.

Step-2)
Compare weighted-cosine similarity scores between stored word sets by frame to determine frame similarity to separate video sequences

Model

• Object Detection with InternImage-XL
  
• Place Recognition with Places365-CNNs