Visual Question Answering (VQA)

Given the input including a picture and a question, the machine will answer the question according to the picture

• trained a LSTM network to encode questions
• trained a CNN to encode images
• fused question and image features into one model and trained the model
• tested on MS COCO dataset

(to be updated, codes created based on https://github.com/GT-Vision-Lab/VQA_LSTM_CNN)