China Offshore SSS-AI: A multi-region side-scan sonar image dataset for cross-regional seafloor target recognition

China Offshore SSS-AI is a multi-region side-scan sonar image dataset designed for seafloor target recognition and cross-regional domain-shift studies. The dataset contains cropped side-scan sonar image chips collected from four representative offshore sectors of China, represented by Dongying, Yantai, Quanzhou and Shenzhen. The release includes image pixels and derived file-level metadata only. Raw sonar records, precise coordinates and survey-line identifiers are not included in order to reduce geospatial and engineering sensitivity. The dataset provides original folder labels, harmonized standard labels, benchmark labels, standardized image names, class mappings, file-level metadata, quality-control records and recommended train/validation/test splits. The dataset can be used for side-scan sonar image classification, cross-region domain-shift analysis, few-shot adaptation, active-learning design and hard-negative evaluation for underwater and seafloor target recognition.