An experiment in computer vision, occupancy-cv collects photos while the computer is being used and while it is not. Then, those images are used to train a system to recognize whether the computer is being used based on an image.
- run collection-vlc.py to gather samples
- run convert.py to prepare the samples
- run train.py to train and run the neural network
- VLC
- python3 (3.2+)
- python-xlib
- tensorflow
- Pillow
- devilspie (to auto-minimize VLC)
- Photo capture
- Training
- Save training data
- Smarter image capture (delta-based, maybe motion?)
- Support capture methods besides VLC
- Support Windows