Lunar Rocks and Craters Dataset for Instance Segmentation (YOLO and COCO Format)

Description

This dataset is designed for the instance segmentation of lunar surface features, specifically rocks and craters, using the YOLO (You Only Look Once) and COCO annotation formats. It provides high-quality annotations tailored for machine learning and computer vision tasks aimed at understanding and analyzing lunar terrains.

Classes: Two primary classes are labeled:

Rocks: Varying in size and shape, representing individual geological features.

Craters: Circular or elliptical depressions of various scales, crucial for lunar surface analysis.

Dataset Composition:

Training Set: 7,734 images 

Testing Set: 307 

Validation Set: 313 images

Image Resolution: All images are resized to 640x640 pixels, maintaining high fidelity while optimizing computational efficiency.

Note: All the data has been captured on the premises of German Research Center for Artificial Intelligence, Robotics Innovation Center