- Author: Andrei Alexandru
- Faculty: Faculty of Electronics, Telecommunications and Information Technology
- Master program: TAID
This code implements a trainer for a multi-task network which performs gaze estimation and emotion classification. It was developed for my Dissertation project.
The hyperparams for an experiment are available and can be modified using the config from configs/efficientnet_train.yaml.
After the desired params are set, one has to run script python tools/train.py.