NitishMutha / paintings-emotion-classifier

Affective Computing and Human Robot Interaction project to identify emotions in paintings using transfer learning

Affective Computing and Human Robot Interaction

Nitish Mutha | Russel Daries | Rafiel Faruq | Charles Olivier | Lewis Moffat | Hugo Philion

Code for Affective Computing and Human Robot Interaction Project

This repository contains:

1. The inception pretrained model in /inception
2. The VGG pretrained caffe model in /Notebooks
3. Our own tensorflow CNNs, in baseline_cnn, deep_cnn and deepest_cnn
4. The data in /data

Other files include helper functions, preprocessing and feature selection. Selective results are included, however, saving each trained model cost > 300mb, so was unfeasible given the limit on Github.

With thanks to Nadia and Nick for the course.