Project Overview

Using manually labeled images from DESI(Dark Energy Spectroscopic Instrument), I trained and optimized a CNN model to automatically identify the rotational direction(clockwise/ counterclockwise) of spiral galaxies, achieving a recognition accuracy of 97.40%. This provides a solution for morphological classification within the context of astronomical big data.

Device on robot

The CNN model training accuracy

Code Repository

🔗 Galaxy rotation classifier — Visit this GitHub repository for full source code.