It solves the problem of manually curating and annotating butterfly wing datasets by offering an interactive web UI built with Streamlit. Users can select the odd wing, record visual reasons, and rate annotations, while the app stores data in Supabase and pulls images from a Hugging Face collection. Designed for researchers and citizen scientists needing a lightweight annotation platform. It stands out by integrating data fetching, storage, and a simple UI without requiring heavy infrastructure.
View on GitHub →kreshuklab/diff_detect