Skip to content
/ vqa Public

This is a CS6240 project on visual question answering.

Notifications You must be signed in to change notification settings

slewyh/vqa

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

8 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Generative Question Answering for Image and Video QA.

GQA, a generative method proposed by Lewis et al., is shown to perform well for image and text data. Our paper aims to extend GQA in 2 ways:

1, Incorporate a generative answer model in for Image-based GQA in order to expand the choices of candidate answers for a question by introducing a new seq2seq model for the answer generator that takes the image and some ‘weak representation’ of the question as inputs.

To view the sub-module and its contents, run:

cd  ImageQA
ls

See README.md for instructions on how to run the model scripts.

2, Apply GQA to video QA on questions that requires spatial and temporal grounding to obtain a relevant answer. The model is evaluated on TVQAplus dataset.

cd  TVQAplus
ls

See README.md for instructions on how to run the model scripts.

About

This is a CS6240 project on visual question answering.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published