galvanized / occupancy-cv

Use a neural network to detect if you're at your computer from a webcam image

occupancy-cv

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.

How to use:

1. run collection-vlc.py to gather samples
2. run convert.py to prepare the samples
3. run train.py to train and run the neural network

Software required:

• VLC
• python3 (3.2+)
• python-xlib
• tensorflow
• Pillow

Software recommended:

• devilspie (to auto-minimize VLC)

Progress:

• Photo capture
• Training
• Save training data
• Smarter image capture (delta-based, maybe motion?)
• Support capture methods besides VLC
• Support Windows