gitaskhub

QMemBN+ is my unified test-time adaptation framework for small vision models that combines quantile-based BN, memory-augmented statistics, and few-shot prototype guidance to adapt on streaming data. It targets real-world distribution shifts with fast, lightweight updates, improving accuracy without retraining or labels.

Stars · 2
Language · Jupyter Notebook
License · MIT
Ask anything about this repo to start.

By chatting or signing in you agree to the Terms and chat-message logging (revocable in History).