Field Crop and Weed Segmentation Dataset

The current agricultural industry faces challenges in managing crops and weeds. Traditional manual identification is inefficient and prone to errors, leading to a decrease in crop yield. Existing solutions rely heavily on manual intervention, lacking efficient automated tools to meet the fine management needs of modern agriculture. This dataset aims to support the training of deep learning models by providing high-quality crop and weed semantic segmentation data, thereby improving the automation level of crop identification and weed monitoring. Data collection uses high-resolution cameras to capture field images under various conditions, ensuring diversity and representativeness. For quality control, we implemented multiple rounds of annotation and consistency checks and invited agricultural experts for review to ensure annotation accuracy and consistency. Data will be stored in JPG format for images and JSON format for label information, facilitating subsequent analysis and use.