Loofah_pests_diseases_image

This study introduces a self-constructed loofah pest and disease dataset comprising 2,929 images collected from real agricultural environments in Guangdong, China. The dataset covers five distinct categories: Downy Mildew (475 images), Diaphania Indica (550 images), Healthy loofah (732 images), Liriomyza (496 images), and Needle Peak (676 images). Collected under diverse lighting conditions and complex field backgrounds using various imaging devices including digital cameras and smartphones, the dataset presents substantial challenges for recognition tasks due to uneven illumination, varied disease severity, and complex environmental contexts. This dataset serves as a valuable benchmark for evaluating lightweight models in real-world agricultural scenarios and supports the development of efficient crop disease diagnosis systems for UAV edge devices.