IP102

The IP102 dataset is a large-scale(fine-grained) benchmark dataset for pest classification, which is primarily composed of images collected from various search engines and entomology-related websites. The dataset encompasses 102 distinct crop pest categories, with a total of 75,222 images. The image scenes in the dataset are realistic, with complex and varied backgrounds, varying lighting conditions, diverse shooting angles, and some images containing noise or watermarks, highly reproducing the actual monitoring environment in the field.