Below are select output images from model-5, which was trained on images from chauvet cave. The code for the model and training images are available in the documentation and the bulk output images can be viewed here. More notes can be found here.
