JeonJaeHyeong/DPL

Code for paper "Semantic Diversity-aware Prototype-based Learning for Unbiased Scene Graph Generation (ECCV 2024)"

25
/ 100
Experimental

This project helps researchers and practitioners in computer vision create more accurate descriptions of relationships between objects within images, known as scene graphs. It takes in collections of images (datasets like VG or GQA) and outputs models that can identify objects and their interactions, such as "person riding bicycle" or "cat on couch," with improved fairness across different types of relationships. Computer vision engineers and AI researchers who develop systems for image understanding or visual question answering would use this.

No commits in the last 6 months.

Use this if you need to generate scene graphs from images and want to ensure that your models are not biased towards common relationships, providing a more balanced and accurate understanding of visual scenes.

Not ideal if you are looking for a pre-trained, ready-to-use model for general image classification or object detection without specific focus on complex relationship prediction or bias mitigation.

computer-vision scene-graph-generation image-understanding AI-research relationship-extraction
Stale 6m No Package No Dependents
Maintenance 2 / 25
Adoption 7 / 25
Maturity 16 / 25
Community 0 / 25

How are scores calculated?

Stars

26

Forks

Language

Jupyter Notebook

License

Last pushed

Jul 23, 2025

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/JeonJaeHyeong/DPL"

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