Saliency Prediction

A project for Cognitive Science 3, 2021/22

The pipeline is controlled through the main.py file.

Instructions for running the code locally

1. After cloning the repo, create a new conda env using conda create -n tf-gpu tensorflow-gpu
2. Install the dependencies from requirements.txt
3. Download the mask_rcnn_coco.h5 from https://github.com/matterport/Mask_RCNN/releases/download/v2.1/mask_rcnn_balloon.h5 and place it inside the models/ folder
4. Specify path to .txt file containing paths to input images in main.py line 3
5. Specify path to image folder in main.py line 4
6. Create folder called 'imageSegmentationOut/'
7. Run the main.py
8. The compressed images will appear in the imageSegmentationOut directory