The actual tool for this project works, just run main.py with your fencing video file specified and BAM, you get the same video with the non-fencing frames removed. On average, this reduces watch time by around 40%, allowing athletes and coaches to streamline their analyzation process.
I did not and do not plan on finishing the website part because of the lack of benefit. Unless this project recieves a ton of support, users will need to download the code to run.
The project uses Google's Tensorflow to create a sophisticated convolutional neural network (CNN), and the OpenCV (computer vision) 2 library to processes video files.