Kangningthu/ADELE

Adaptive Early-Learning Correction for Segmentation from Noisy Annotations (CVPR 2022 Oral)

40
/ 100
Emerging

When analyzing images to identify specific objects or regions, you often have to rely on existing, imperfect labels. This tool helps improve the accuracy of automatically generated segmentation masks even when your initial training data annotations are noisy or incomplete. It takes your images and their preliminary, potentially flawed segmentation masks as input and refines these masks to produce more precise segmentation boundaries. This is useful for researchers or practitioners working with image analysis where precise object delineation is critical, but perfect manual labeling is impractical.

No commits in the last 6 months.

Use this if you need to accurately identify and separate objects within images, but your available segmentation annotations are imperfect, noisy, or were generated semi-automatically.

Not ideal if you have perfectly clean, manually verified segmentation masks for all your images, or if your primary goal is object detection rather than precise pixel-level delineation.

medical-imaging image-analysis computer-vision data-labeling annotation-refinement
Stale 6m No Package No Dependents
Maintenance 0 / 25
Adoption 9 / 25
Maturity 16 / 25
Community 15 / 25

How are scores calculated?

Stars

89

Forks

12

Language

Python

License

MIT

Last pushed

Jun 19, 2022

Commits (30d)

0

Get this data via API

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

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