jiwoon-ahn/irn

Weakly Supervised Learning of Instance Segmentation with Inter-pixel Relations, CVPR 2019 (Oral)

49
/ 100
Emerging

This project helps computer vision practitioners automatically identify and outline individual objects within an image, even when only broad category labels are available for the entire image. It takes an image along with a general class label (e.g., "cat" for an image containing a cat) and outputs precise, pixel-level outlines for each instance of that object. Researchers and engineers working with image analysis benefit from this for tasks requiring detailed object localization.

535 stars. No commits in the last 6 months.

Use this if you need to precisely segment individual objects in images but only have coarse, image-level class labels, not detailed pixel-level annotations.

Not ideal if you already have fully annotated datasets with precise pixel masks for all objects or if your task only requires simple bounding box detection.

image-segmentation object-detection computer-vision-research weakly-supervised-learning image-analysis
Stale 6m No Package No Dependents
Maintenance 0 / 25
Adoption 10 / 25
Maturity 16 / 25
Community 23 / 25

How are scores calculated?

Stars

535

Forks

96

Language

Python

License

MIT

Last pushed

May 01, 2023

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/computer-vision/jiwoon-ahn/irn"

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