Repository files navigation

This repository contains the implementation to develop a gaze estimator for 128x128 full face images.It also contains the error analysis performed on the predictions of the same.

This code was developed and tested on Ubuntu 18.04.5 LTS, it requires GPUs andCUDA support.The python version used was 3.8.0 .To run the code you would needJupyter Software. You can also configurecustom kernels for it in VS code.

All other requirements can be installed with:pip install -r requirements.txt

GazeCapture.h5, MPIIGaze.h5 are the training and testing sets used respectively. To obtain these please follow steps mentionedhere to pre process the original datasets.

Redirected_samples1.h5, is the augmented dataset please usethis to generate the augmented dataset.

For doing the error analysis, the predictions are loaded fromloss_2303196.pkl pickle file. Please use the code intrain_gaze_estimator.ipynb to generate predictions on the augmented dataset.

The notebooktrain_gaze_estimation.ipynb contains the code to train/ test the model, the cells are organized so that one may perform both separately.The resultant plots are saved inplots folder.

The notebookerror_analysis.ipynb contains the code to perform error analysis on the predictions.

• Model trained on Augmented Dataset, for a total of 20k Images and 20 epochs

• Paper
• Gaze Redirection Code Repository
• Gaze Capture Dataset
• MPIIGaze Dataset