Closed ggsonic closed 4 years ago
Link to #155
Awesome! I'll try it today.
BTW: I'm thinking about if we can use some image feature extracting model to get the image feature from the video every second, and then use a bi-lstm(or transformer etc) to model the start point and the end point.
Just a mind storm...
add video qrcode cut script