This repository contains the code for feature extraction, treaining, testing and the user interface for the paper "Expert Knowledge-Aware Image Difference Graph Representation Learning for Difference-Aware Medical Visual Question Answering". (Project page)
Screen.Recording.2023-07-29.at.02.04.27.mov
The Medical-Diff-VQA (originally MIMIC-Diff-VQA) dataset can be downloaded from Physionet (currently under review. will be available soon). The code used for generating the dataset can be found in the Medical-Diff-VQA repository.
This repository contains two main folders. The feature extraction folder contains the code for extracting the image features. The model folder contains the code for training and testing our model. Please refer to the README.md in each folder for more details.