autoBleep
A neural net enabled video project to read lips and anticipate bad words
The project
The AutoBleep is a program designed to bleep live video/audio, using video trained on a subjects face. The objective is for the program to detect the first part of a bad word though live video analysis.
The tech
The project is built off of openCV and keras primarily. An openCV program running a haar classification cascade grabs pictures of a subjects mouth every few frames, then runs them through the classifcation model.
Built With
- keras
- python
- sklearn
Log in or sign up for Devpost to join the conversation.