Requirements

Primary requirements to execute:

1. python > 3.5
   
2. pytorch > 1.0.1
   

Bottom-Up and Top-Down Attention for Visual Question Answering

This repository uses Bottom-Up and Top-Down Attention for Visual Question Answering visual features. These features are extracted for top-k salient objects detected through Faster-RCNN.

Instructions

1. Execute as: python3 run.py
   
2. After execution all models and results log will be saved in 'result/' directory.
   
3. All data should be placed inside 'data/' directory.