Starter code will be provided in due course time

Example codes are provided to give an idea about the workflow

1) GazeRegularizedAttention.py - contains the code for building the gaze regularized attention mechanism
    along with calculating the gaze distribution for an image
2) Flamingo.py - provides the modifications made to Flamingo.py and incorporation of gaze-regularized attention mechanism
3) CreateModel.py - deals with the actual creation of the model where we make the gaze regularized attention mechanism trainable 
    also we provide code for how to deal with FSDP
4) train_gaze.py - provides code for how training is done using the regularization loss and cross entropy loss
5) gaze_overlay.py- provides code for how gaze is overlaid on images after a heatmap is created 
6) opticalflow_gazecorrection.py - provides a vague code for how the process of occlusion check is performed using RAFT model 