bknyaz/sgg
Train Scene Graph Generation for Visual Genome and GQA in PyTorch >= 1.2 with improved zero and few-shot generalization [BMVC 2020, ICCV 2021]
This project helps computer vision researchers and practitioners accurately interpret complex visual scenes by identifying objects and the relationships between them. It takes in images with detected objects and generates a 'scene graph' describing these relationships, even for rare or unseen combinations. This is especially useful for those working on advanced image understanding and visual reasoning tasks.
141 stars. No commits in the last 6 months.
Use this if you need to build or evaluate models that can understand not just 'what' is in an image, but also 'who is doing what to whom' or 'how things relate to each other' for various computer vision applications.
Not ideal if you are looking for a pre-trained, production-ready system for basic object detection or image classification tasks without delving into object relationships.
Stars
141
Forks
20
Language
Jupyter Notebook
License
—
Category
Last pushed
Jun 18, 2023
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/diffusion/bknyaz/sgg"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.