MCD-Net: A Large-scale Optical-Only Moraine Segmentation Dataset

This dataset comprises the first large-scale, optical-only moraine segmentation collection, specifically designed for advancing automated glacial geomorphological mapping. The dataset includes: 1. Primary Moraine Segmentation Dataset (2,900+ image-mask pairs): - High-resolution optical imagery from Google Earth Pro (0.5-2 meters/pixel) - Binary segmentation masks distinguishing moraine bodies from background - Coverage across key glacial regions in Sichuan and Yunnan, China - Diverse moraine types including lateral, terminal, and recessional moraines 2. Supplementary Ridge Annotation Dataset (400+ image-mask pairs): - Including moraine ridge features - Provides additional morphological information for advanced analysis - Useful for studying fine-scale moraine characteristics The dataset supports the development and benchmarking of deep learning models for moraine identification, particularly focusing on optical-only approaches that eliminate dependency on digital elevation models (DEMs). This resource enables reproducible research in cryospheric remote sensing and facilitates the development of scalable monitoring tools for climate change impacts in high-altitude glacial environments.