Soil Texture and Particle Segmentation Dataset

The current agricultural sector faces challenges in inadequate soil quality monitoring. Traditional soil analysis methods are often time-consuming and costly, making it difficult to meet the needs of precision agriculture. Existing solutions often lack sufficient precision and real-time capabilities, unable to effectively support agricultural decision-making. This dataset aims to provide high-quality semantic segmentation data on soil texture and particles to support automated image analysis technologies, enhancing the efficiency and accuracy of soil monitoring. Data collection is conducted using high-resolution cameras in different soil environments to ensure coverage of various soil types. In terms of quality control, we implemented multiple rounds of annotation and consistency checks, with annotation results reviewed by domain experts. Data is stored in JPG and JSON formats, organized in such a way that each image corresponds to one record, containing image data and its label information.