Image Region Features - Githubissues

facebookresearch / mmf

A modular framework for vision & language multimodal research from Facebook AI Research (FAIR)

https://mmf.sh/

Other

5.49k stars 935 forks source link

Image Region Features #1105

Closed harshraj22 closed 2 years ago

harshraj22 commented 3 years ago

❓ Questions and Help

Hello, I would like to use the repo's code, which extracts the image's region features (as bottom up attention) in my personal project. I would like to add it to my architecture such that it is end to end trainable. Can someone guide me which file/ class to take ?

I went through the official documentation, but it seems, this would not make the architecture end to end trainable.

TownWilliam commented 2 years ago

Hello, it seems that the python scripts extracting the features is in the location : mmf/tools/scripts/features/extract_features_vmb.py.

This script provides several params and Resnet checkpoint files to get the wanted features results. It could covert the original image files into corresponding region features.