Orchard Scene Recognition Dataset

The current agricultural industry faces issues with low orchard management efficiency and untimely monitoring of pests and diseases. Existing solutions offer a degree of monitoring but often lack real-time capabilities and precision, preventing farmers from reacting promptly. This dataset aims to enhance the application of object detection algorithms in orchard management by providing high-quality orchard scene images and annotations. Data collection is conducted using drones, covering different seasons and lighting conditions to ensure data diversity. We implemented multiple rounds of annotation and expert reviews to ensure annotation quality. In addition, the data is stored in JPEG format and accompanied by JSON files providing annotation information for easy use.