SakurajimaMaiii/TSD

[CVPR 2023] Feature Alignment and Uniformity for Test Time Adaptation

33
/ 100
Emerging

This tool helps machine learning engineers and researchers improve the reliability of their image classification models when applied to new, unseen environments. It takes a trained image recognition model and new image data from a different domain, then adapts the model to maintain high accuracy on this new data without needing to retrain on the original dataset. This is particularly useful for those deploying models in dynamic real-world settings where data characteristics can shift.

Use this if your existing image classification model's performance drops when encountering images from a new visual environment or style, and you need to adapt it quickly without access to the original training data.

Not ideal if you are looking for a general-purpose machine learning library or if your problem does not involve adapting pre-trained image models to new visual domains.

image-classification domain-adaptation computer-vision model-deployment machine-learning-research
No Package No Dependents
Maintenance 6 / 25
Adoption 8 / 25
Maturity 16 / 25
Community 3 / 25

How are scores calculated?

Stars

48

Forks

1

Language

Python

License

MIT

Last pushed

Nov 23, 2025

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/diffusion/SakurajimaMaiii/TSD"

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