coldmanck/recovering-unbiased-scene-graphs

Official implementation of "Recovering the Unbiased Scene Graphs from the Biased Ones" (ACMMM 2021)

25
/ 100
Experimental

This project helps computer vision researchers and practitioners create more accurate and balanced scene graphs from images. It takes raw images as input and produces graphical representations that describe visual relationships between objects, while specifically addressing "reporting bias" where some relationships are underrepresented. Researchers working on image understanding, visual question answering, or autonomous systems would find this useful for improving their models' perception of visual scenes.

No commits in the last 6 months.

Use this if you are working with scene graph generation and your models are biased towards common relationships, missing many valid but less frequent object interactions.

Not ideal if you are looking for a general-purpose image labeling tool or if you do not have significant expertise in computer vision and deep learning.

Computer Vision Image Understanding Scene Graph Generation Visual AI Model Debiasing
No License Stale 6m No Package No Dependents
Maintenance 0 / 25
Adoption 9 / 25
Maturity 8 / 25
Community 8 / 25

How are scores calculated?

Stars

78

Forks

5

Language

Python

License

Last pushed

Sep 04, 2022

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/coldmanck/recovering-unbiased-scene-graphs"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.