sarthaxxxxx/BATCLIP

[ICCV '25] BATCLIP: Bimodal Online Test-Time Adaptation for CLIP

35
/ 100
Emerging

This tool helps computer vision researchers and AI practitioners improve how well image classification models, like CLIP, handle real-world images that might be blurry, noisy, or corrupted. It takes in unlabeled, corrupted images and a pre-trained CLIP model, then adapts the model to make its classifications more accurate. The output is a more robust image classification model, better able to understand images in challenging conditions.

No commits in the last 6 months.

Use this if you are working with open-vocabulary image classification and need to make your models more reliable when encountering corrupted or 'noisy' images in real-time.

Not ideal if your primary concern is training models from scratch or if your image data is consistently clean and free from corruption.

computer-vision image-classification model-robustness AI-research online-learning
Stale 6m No Package No Dependents
Maintenance 2 / 25
Adoption 5 / 25
Maturity 15 / 25
Community 13 / 25

How are scores calculated?

Stars

9

Forks

2

Language

Python

License

MIT

Last pushed

Jul 03, 2025

Commits (30d)

0

Get this data via API

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

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