Automatic Detection of Craters and Boulders from Orbiter High-Resolution Camera (OHRC) Images Using YOLO Models

This work investigates the automated detection of lunar craters and boulders using advanced Computer Vision (CV) techniques, specifically leveraging the YOLOv11 and YOLO-NAS object detection models on high-resolution images from ISRO’s Chandrayaan-2 Orbiter High-Resolution Camera (OHRC). Recognizing the importance of lunar crater studies for understanding impact history, geological evolution, and planning future missions, this research aims to develop a more efficient and scalable solution for identifying and analyzing lunar surface features. The primary objectives include training the YOLOv11 and YOLO-NAS models on OHRC data to detect craters and boulders, evaluating model accuracy using metrics such as mean Average Precision (mAP), and refining detection parameters to optimize performance.