River Ice Fine-Grained Segmentation: A GF-2 Satellite Image Dataset and Deep Learning Benchmark

Semantic segmentation of river ice images serves as a critical technological foundation for hydrological monitoring and an ice flood early warning system. Current publicly available river ice datasets predominantly utilize UAV-captured images and ground-based photographic observations. To address the limitations of spatial coverage in existing datasets, we present NWPU_YRCC_GFICE—a satellite remote sensing dataset constructed from multispectral GF-2 satellite images. The dataset innovatively categorizes river ice into six fine-grained classes across freeze–thaw cycles and covers river ice data from the Yellow River (Ningxia-Inner Mongolia section) spanning the past ten years. We further establish a comprehensive deep learning benchmark, which evaluates 33 state-of-the-art segmentation models and two improved segmentation models based on YOLO and SegFormer architectures, separately. Experiments are conducted on the NWPU_YRCC_GFICE dataset and three public river ice datasets (NWPU_YRCC_EX, NWPU_YRCC2, and Alberta river ice segmentation datasets). The proposed models exhibit excellent performance, surpassing the state-of-the-art methods. The presented NWPU_YRCC_GFICE dataset and the benchmark enrich the river ice dataset and favor promoting fine-grained river ice segmentation research from satellite view.