mashijie1028/Happy-CGCD

(NeurIPS 2024) Happy: A Debiased Learning Framework for Continual Generalized Category Discovery

39
/ 100
Emerging

This framework helps machine learning practitioners continually discover new categories in incoming, unlabeled image data streams, even when those streams might contain previously seen items. It takes a sequence of image datasets, some partially labeled for initial training and subsequent ones entirely unlabeled, and outputs a continually updated model capable of classifying both known and newly discovered categories. This is ideal for researchers and practitioners working on dynamic image classification systems.

Use this if you need to build a system that can learn new visual categories over time from unlabeled data, without forgetting old ones, in an ongoing, unsupervised manner.

Not ideal if your image classification task involves only a fixed set of known categories or if you have fully labeled data for all new classes that appear.

continual-learning unsupervised-learning image-categorization novelty-detection machine-vision
No Package No Dependents
Maintenance 6 / 25
Adoption 8 / 25
Maturity 16 / 25
Community 9 / 25

How are scores calculated?

Stars

45

Forks

4

Language

Python

License

MIT

Last pushed

Nov 25, 2025

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/mashijie1028/Happy-CGCD"

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