Unsupervised learning to find facial features. https://youtu.be/4VAkrUNLKSo
The main order is:
- edit
datagen.py
to point to the correct directory, etc. - run
python datagen.py
- (this creates a bunch of training data)
- run
python encoder.py
- This starts modeling the data, and periodically saves off evaluation summaries as well as the all-important
Encoder.h5
- Once you have an
Encoder.h5
, runface_edit.py
to see the result of applying different dimensions.