Research on Target Detection of Fruit and Vegetable Diseases Based on Td-Fd-YOLOv8 Algorithm

Recent advancements in deep learning technologies have significantly contributed to various fields of computer vision. In the context of fruit and vegetable disease detection, challenges such as the difficulty of detecting small object features, low accuracy, and inefficient performance of traditional algorithms remain prevalent. To address these issues, this paper presents an improved Td-Fd object detection algorithm based on YOLOv8. The proposed algorithm builds upon the YOLOv8 model and optimizes the backbone network by incorporating the EMBC module, thereby enhancing its feature extraction capability. The FPSC module replaces the original SPPF layer to improve the efficiency of multi-scale feature capture. Additionally, the CCFM module is integrated into the neck network to strengthen the fusion of features at different scales. The experiments were conducted using the Roboflow Cantaloupe Disease dataset, which includes disease categories such as aphids, leafminers, moths, red melon beetles, and whiteflies. Experimental results demonstrate that the proposed algorithm significantly improves both the accuracy and efficiency of small target detection, with an overall average detection accuracy reaching 84.8 %, providing technical support for the practical application of fruit and vegetable disease detection.