Skip to content

apugoneappu/ask_me_anything

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Ask Me Anything: A tool for visualising Visual Question Answering (AMA) Open in Streamlit

An easy-to-use app to visualise attentions of various VQA models. Please click here to see a live demo of the app!

top 7 predictions

Models
Requirements
Installation
How to run
How to use
Contributing
• Acknowledgements

Models

• MFB - Multi-modal Factorized Bilinear Pooling with Co-Attention Learning for Visual Question Answering
Zhou Yu, Jun Yu, Jianping Fan, Dacheng Tao
Arxiv

• (Coming soon) MCAN - Deep Modular Co-Attention Networks for Visual Question Answering
Zhou Yu, Jun Yu, Yuhao Cui, Dacheng Tao, Qi Tian
Arvix

Requirements

Please check the requirements.txt file for the version numbers.

  1. opencv_python==4.4.0.46
  2. numpy==1.19.4
  3. pandas==1.1.4
  4. torch==1.4.0
  5. matplotlib==3.3.2
  6. gdown==3.12.2
  7. seaborn==0.11.0
  8. dotmap==1.3.23
  9. streamlit==0.70.0
  10. Pillow==8.0.1
  11. PyYAML==5.3.1

Installation

  1. Install Anaconda
  2. Clone this repository and cd into it.
    git clone https://github.com/apugoneappu/ask_me_anything.git && cd ask_me_anything
  3. In a new environment (new_env)
    pip install -r requirements.txt

How to run

From the directory of this repository, do the following -

  1. conda activate new_env
  2. streamlit run main.py
  3. In a browser tab, open the Network URL displayed in your terminal.

Done! 🎉

How to use

input page image attentions text attentions

Contributing

First of all, thank you for wanting to contribute to this work! I will try and make your job as easy as possible. Detailed instructions coming soon ...

Acknowledgements

This repository has been built by modifying the OpenVQA repository.

I would also like to thank Yash Khandelwal, Nikhil Shah and Chinmay Singh for their support and amazing suggestions!

Huge thanks to Streamlit for making all of this possible and for Streamlit Sharing that enables free hosting of this app! ❤️

About

An easy-to-use app to visualise attentions of various VQA models.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages