Durian Plantation Pest Psyllid Identification Image Data

In recent years, durian plantations have faced serious pest psyllid problems, affecting yields and increasing pesticide usage frequency. Existing manual identification methods are inefficient and have a high rate of misjudgment, which is not conducive to large-scale plantation management. This dataset, annotated with high-quality images, aims to solve the problem of intelligent pest identification, improving accuracy and efficiency. Data collection is done via high-definition cameras and drones in the fields, covering pest manifestations under different seasons and environments. The data quality is ensured through multiple rounds of expert annotation and consistency checks, with the annotation team comprised of plant pathology experts and computer vision experts, with a team size of 30 people. Image preprocessing uses techniques like brightness adjustment and noise filtering. Final data is stored in JPG format, organized with accompanying label files. The annotation precision of images in the dataset reaches 95%, is comprehensive, and covers pest manifestations under different growth cycles and environments. Technological innovation lies in using deep learning algorithms to optimize the annotation process, enhancing data annotation efficiency. The application value is reflected in improving the accuracy of durian pest identification and significantly reducing pest prevention costs. Compared to other agricultural datasets, this dataset has higher specificity and applicability, especially addressing durian plantation issues in Southeast Asia, it is scarce and has expandable application space, applicable to other similar pest and disease identification tasks.