yashgupta-7/visual-question-answering

Visual Question Answering using CNN-LSTM based networks

members:
Raaghav Raaj (180050082)
Vrinda Jindal (180050120)
Yash Gupta (180050121)

project title:
Visual Question Answering

./Code files:
models.py - keras implementation of models 1,3 and 4 (no model 2)
vqa_data_stats.ipynb - study various statistics of the dataset
vqa_training.ipynb - train models 1 and 3
vqa_training_vis.ipynb - train model 4
vqa_see_image.ipynb - visualise the image, given the dataset
vqa_predict.ipynb - predict given the model and standard/custom question


./Code downloads:
fill data/, trained_models/ and /nlp_vqa according to the .txt files given.