农业装备与车辆工程2025，Vol.63Issue(9)：13-18,6.DOI:10.3969/j.issn.1673-3142.2025.09.003
基于改进YOLOv7的苹果目标检测算法研究
Research on apple object detection algorithm based on improved YOLOv7
摘要
Abstract
In response to the problems of insufficient adaptability and high computational complexity of apple fruit detection models in natural scenes,improvements was made based on YOLOv7,and an ESS-YOLO model that balanced accuracy and inference speed was proposed.Firstly,the improved SPPFCSPC module was used to replace the traditional SPPCSPC module in YOLOv7,thereby reducing the computational load and parameter quantity of the model.Secondly,the ECA attention mechanism module was introduced to enhance the robustness and generalization ability of the model.Finally,the SIoU loss function was adopted to accelerate the convergence speed of the model.Experimental results show that the precision,recall,and average precision of ESS-YOLO reach 92.3%,93.9%,and 99.4%respectively.Compared with the benchmark model YOLOv7,the precision was increased by 1.1%,the recall was increased by 0.8%,and the average precision was increased by 1.8%.Meanwhile,its parameter quantity and floating-point computation were reduced by 0.12 M and 2.3 G respectively,and the inference speed FPS was improved by 18%.The ESS-YOLO model demonstrated excellent adaptability and robustness,which could provide technical references for promoting the application of intelligent apple picking.关键词
苹果/YOLOv7/目标检测/注意力机制Key words
apple/YOLOv7/target detection/attention mechanism分类
农业科技引用本文复制引用
杨帆,周杰,吴昊荣,罗瑶,汪旭,廖佳豪..基于改进YOLOv7的苹果目标检测算法研究[J].农业装备与车辆工程,2025,63(9):13-18,6.基金项目
四川省大学生创业训练计划项目"视觉检测工业应用系统研究"(S202411079012X) （S202411079012X）
