Objective In the context of large-scale apple orchard planting and management, traditional orchard inspections are prone to false detections and omissions. Additionally,traditional apple disease detection models have a large number of parameters, making it different to deploy them on mobile devices. Therefore, designing an efficient and lightweight apple disease detection model can enable effective prevention and precise management of apple diseases, thereby improving apple quality and increasing orchard economic income. Methods To address these needs, an improved algorithm based on YOLOv5s is proposed. The common diseases anthracnose and brown spot are the primary research subjects. Apple skin disease images were collected to construct an orchard apple disease dataset, and images were annotated and classified using the Labelimg tool. The GhostNet lightweight module was introduced to replace the main feature extraction network, capturing more feature information with fewer parameters to achieve a lightweight model suitable for mobile deployment. The SimAM parameter-free attention mechanism was introduced to strengthen the model's simultaneous focus on channel and spatial information, assigning higher priority to important disease features without adding any parameters, thus improving model accuracy. The SIoU boundary box regression loss function was introduced to optimize the accurate localization of the predicted box to the target diseases, quickly locating the accurate axis by redefining the angle penalty metric and using a genetic algorithm to optimize the value of θ, thereby enhancing model training and inference capabilities. Results The improved model’s parameters and floating-point operations (FLOPs) were reduced by 30.2% and 33.8%, respectively, compared to the original model. On the basis of achieving lightweight performance, the mAP@0.5 reached 93.7%, and the mAP@0.5:0.95 reached 63.3%, which were 1.6% and 0.7% better than the original YOLOv5s algorithm, respectively. Conclusion The improved model achieves good detection performance while maintaining a lightweight design, efficiently identifying apple diseases and providing technical support and reference for the detection of diseases in other crops.
BalasubramaniamV .Artificial Intelligence Algorithm with SVM Classification using Dermascopic Images for Melanoma Diagnosis[J].Journal of Artificial Intelligence and Capsule Networks,2021,3(1):34-42.
[2]
GirshickR .Fast R-CNN[C].2015 IEEE International Conference on Computer Vision (ICCV).Santiago:IEEE,2015:1440-1448.
WangC Y, BochkovskiyA, LiaoH Y M .Scaled-YOLOv4:scaling cross stage partial network[C].2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).Nashville:IEEE,2021:13024-13033.
[5]
WangC Y, BochkovskiyA, LiaoH Y M .YOLOv7:trainable bag-of-freebies sets new state-of-the-art for real-time object detectors[C].2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).Vancouver:IEEE,2023:7464-7475.
[6]
YangZ T, SunY N, LiuS,et al .3DSSD:point-based 3D single stage object detector[C].2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).Seattle:IEEE,2020:11037-11045.
[7]
ZhengW, TangW L, JiangL,et al .SE-SSD:self-ensembling single-stage object detector from point cloud[C].2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).Nashville:IEEE,2021:14489-14498.
[8]
ChakrabortyS, PaulS, Rahat-uz-ZamanM .Prediction of apple leaf diseases using multiclass support vector machine[C].2021 2nd International Conference on Robotics,Electrical and Signal Processing Techniques (ICREST).Dhaka:IEEE,2021:147-151.
YangY R, WuH R, ZhangY,et al .Tomato disease recognition using leaf image based on complex environment[J].Journal of Chinese Agricultural Mechanization,2021,42(9):177-186.
XueY, WangL Y, ZhangY,et al .Defect detection method of apples based on GoogLeNet deep transfer learning[J].Transactions of the Chinese Society for Agricultural Machinery,2020,51(7):30-35.
XiongM Y, ZhanW, GuiL Y,et al .Detection and identification of corn leaf disease based on ResNet model[J].Jiangsu Agricultural Sciences,2023,51(8):164-170.
YeS F, ShiZ H, SuC Y,et al .Research on real-time detection algorithm of power line and pole tower based on YOLOv5[J].Computer Measurement & Control,2022,30(11):77-84.
[19]
GhiasiG, LinT Y, LeQ V .NAS-FPN:learning scalable feature pyramid architecture for object detection[C].2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).Long Beach:IEEE,2019:7029-7038.
[20]
ZhouL M, RaoX H, LiY H,et al .A lightweight object detection method in aerial images based on dense feature fusion path aggregation network[J].ISPRS International Journal of Geo-Information,2022,11(3):189.
[21]
HanK, WangY H, TianQ,et al .GhostNet:more features from cheap operations[C].2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).Seattle:IEEE,2020:1577-1586.
[22]
YangL X, ZhangR Y, LiL D,et al .Simam:A simple,parameter-free attention module for convolutional neural networks[C].International conference on machine learning.PMLR,2021:11863-11874.
[23]
GevorgyanZ .SIoU loss:more powerful learning for bounding box regression[EB/OL].[2024-04-08].2022:arXiv:2205.12740.org/abs/2205.12740.
[24]
HowardA, SandlerM, ChenB,et al .Searching for MobileNetV3[C].2019 IEEE/CVF International Conference on Computer Vision (ICCV).Seoul:IEEE,2019:1314-1324.
[25]
HuJ, ShenL, SunG .Squeeze-and-excitation networks[C].2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition.Salt Lake City:IEEE,2018:7132-7141.
[26]
WangQ L, WuB G, ZhuP F,et al .ECA-net:efficient channel attention for deep convolutional neural networks[C].2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).Seattle:IEEE,2020:11531-11539.
[27]
WooS, ParkJ, LeeJ Y,et al .CBAM:convolutional block attention module[C].Computer Vision–ECCV 2018.Cham:Springer International Publishing,2018:3-19.
[28]
HouQ B, ZhouD Q, FengJ S .Coordinate attention for efficient mobile network design[C].2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).Nashville:IEEE,2021:13708-13717.
[29]
TongZ J, ChenY H, XuZ W,et al .Wise-IoU:bounding box regression loss with dynamic focusing mechanism[EB/OL].[2024-04-08].2023:arXiv:2301.10051.org/abs/2301.10051.
[30]
LiuW, AnguelovD, ErhanD,et al .SSD:single shot MultiBox detector[C]//Leibe B,Matas J,Sebe N,et al.,eds.Computer Vision–ECCV 2016.Cham:Springer International Publishing,2016:21-37.
[31]
GirshickR .Fast R-CNN[C].2015 IEEE International Conference on Computer Vision (ICCV).Santiago:IEEE,2015:1440-1448.
[32]
GeZ, LiuS T, WangF,et al .YOLOX:exceeding YOLO series in 2021[EB/OL].[2024-04-08].2021:arXiv:2107.08430.org/abs/2107.08430.
ZhaoJ W, TianG Z, QiuC,et al .Detection method of apple leaf diseases based on improved YOLOv4 algorithm[J].Jiangsu Agricultural Sciences,2023,51(9):193-199.
WangQ S, LüL, HuangD F,et al .Research of apple leaf disease defect detection based on improved YOLOv4 algorithm[J].Journal of Chinese Agricultural Mechanization,2022,43(11):182-187.