基于改进YOLOv5的番茄果实识别估产方法

An improved YOLOv5-based method for tomato fruit identification and yield estimation

摘要

为了实现大棚环境中番茄的智能在线产量估算，提出了一种基于改进的YOLOv5(You Only Look Once v5)番茄识别算法，对自然生长状态下的番茄果实产量进行统计和估算。首先，使用可分离视觉转换器(Separable Vision Transformer, SepViT)替换骨干网络的最后一层，以增强骨干网络与全局信息之间的联系并提取番茄特征；其次，引入WIOU (Wise Intersection over Union)损失函数和Mish激活函数，以提高收敛速度和精度。试验结果表明，改进后的检测模型在平均精度(mean Average Precision, mAP)方面达到了99.5%，相较传统的YOLOv5模型提高了1.1个百分点，每张图像的处理时间为15 ms。此外，改进后的YOLOv5算法对密集和遮挡情况下的番茄果实识别效果更好。

Abstract

In pursuit of intelligent real-time yield estimation for tomatoes in greenhouse environments, we introduce an enhanced YOLOv5 tomato recognition algorithm aimed at the statistical assessment and estimation of tomato fruit yield in their natural growth conditions. Our approach involved two key enhancements: firstly, we substituted the final layer of the backbone network with a Separable Vision Transformer to augment the connectivity between the backbone network and global context, thereby facilitating tomato feature extraction; secondly, we incorporated the WIOU loss function and employ the Mish activation function to enhance convergence speed and accuracy. Experimental findings demonstrate that the improved detection model achieves a mAP score of 99.5%, reflecting 1.1 percentage points enhancement compared to the conventional YOLOv5 model, and the processing time for every image is 15ms. Furthermore, the improved YOLOv5 algorithm exhibits superior recognition performance for densely populated and occluded tomato fruits.

关键词

大棚番茄 / YOLOv5 / 注意力机制 / 损失函数

Key words

Greenhouse tomato / YOLOv5 / Attention mechanism / Loss function

参考文献

[1] 张俊宁, 毕泽洋, 闫英, 等. 基于注意力机制与改进YOLO的温室番茄快速识别[J]. 农业机械学报, 2023, 54(5): 236-243.

[2] 孙皓泽, 常天庆, 王全东, 等. 一种基于分层多尺度卷积特征提取的坦克装甲目标图像检测方法[J]. 兵工学报, 2017, 38(9): 1681-1691.

[3] LIU G X, MAO S Y, KIM J H. A mature-tomato detection algorithm using machine learning and color analysis[J]. Sensors, 2019, 19(9): 2023.

[4] 李寒, 陶涵虓, 崔立昊, 等. 基于SOM-K-means算法的番茄果实识别与定位方法[J]. 农业机械学报, 2021, 52(1): 23-29.

[5] MALIK M H, ZHANG T, LI H, et al. Mature tomato fruit detection algorithm based on improved HSV and watershed algorithm[J]. IFAC-Papers On Line, 2018, 51(17): 431-436.

[6] 马翠花, 张学平, 李育涛, 等. 基于显著性检测与改进Hough变换方法识别未成熟番茄[J]. 农业工程学报, 2016, 32(14): 219-226.

[7] GIRSHICK R, DONAHUE J, DARRELL T, et al. Rich feature hierarchies for accurate object detection and semantic segmentation[C]. IEEE Conference on Computer Vision and Pattern Recognition, 2014.

[8] GIRSHICK R. Fast R-CNN[C]. IEEE International Conference on Computer Vision, 2015.

[9] REN S Q, HE K M, GIRSHICK R, et al. Faster R-CNN:Towards real-time object detection with region proposal networks[J]. Advances in Neural Information Processing Systems, 2015.

[10] REDMON J, FARHADI A. YOLO9000:Better, faster, stronger[C]. IEEE Conference on Computer Vision and Pattern Recognition, 2017.

[11] REDMON J, DIVVALA S, GIRSHICK R, et al. You only look once:Unified, real-time object detection[C]. IEEE Conference on Computer Vision and Pattern Recognition, 2016.

[12] LIU W, ANGUELOV D, ERHAN D, et al. Ssd:Single shot multibox detector[C]. Computer Vision-ECCV 2016:14th European Conference, Amsterdam, The Netherlands, October 11-14，2016，Proceedings, Part I 14.Springer International Publishing, 2016.

[13] 何斌, 张亦博, 龚健林, 等. 基于改进YOLO v5的夜间温室番茄果实快速识别[J]. 农业机械学报, 2022, 53(5): 201-208.

[14] 周云成, 许童羽, 邓寒冰, 等. 基于双卷积链Fast R-CNN的番茄关键器官识别方法[J]. 沈阳农业大学学报, 2018, 49(1): 65-74.

[15] 张文静, 赵性祥, 丁睿柔, 等. 基于Faster R-CNN算法的番茄识别检测方法[J]. 山东农业大学学报(自然科学版), 2021, 52(4): 624-630.

[16] 刘芳, 刘玉坤, 林森, 等. 基于改进型YOLO的复杂环境下番茄果实快速识别方法[J]. 农业机械学报, 2020, 51(6): 229-237.

[17] LIU G X, NOUAZE J C, TOUKO P L, et al. YOLO-Tomato:A robust algorithm for tomato detection based on YOLOv3[J]. Sensors-Basel, 2020, 20(7): 2145.

[18] 成伟, 张文爱, 冯青春, 等. 基于改进YOLOv3的温室番茄果实识别估产方法[J]. 中国农机化学报, 2021, 42(4): 176-182.

[19] REZATOFIGHI H, TSOI N, GWAK J Y, et al. Generalized intersection over union:A metric and a loss for bounding box regression[C]. IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2019: 658-666.

[20] 成伟. 基于深度学习的番茄植株多目标识别方法研究[D]. 太原: 太原理工大学, 2021.

[21] VASWANI A, SHAZEER N, PARMAR N, et al. Attention is all you need[C]. Proceedings of the 31st Information Processing Systems, 2017: 6000-6010.

[22] DOSOVITSKIYA, BEYER L, KOLESNIKOV A, et al. An image is worth 16×16 words:Transformers for image recognition at scale[J]. arXiv:2010.11929， 2020.

[23] LI W, WANG X, XIA X, et al. Sepvit:Separable vision transformer[J]. arXiv:2203.15380， 2022.

[24] HOWARD A G, ZHU M L, CHEN B, et al. MobileNets:Efficient convolutional neural networks for mobile vision applications[J]. arXiv:1704.04861， 2017.

[25] SANDLER M, HOWARD A, ZHU M, et al. MobileNetv2:Inverted residuals and linear bottlenecks[C]. IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2018: 4510-4520.

[26] ANDREW H, MARK S, GRACE C, et al. Searching for mobilenetv3[C]. Proceedings of the IEEE/CVF international conference on computer vision, 2019: 1314-1324.

[27] YU J H, JIANG Y N, WANG Z Y, et al. Unitbox:An advanced object detection network[C]. Proceedings of the 24th ACM international conference on Multimedia, 2016: 516-520.

[28] ZHENG Z H, WANG P, REN D W, et al. Enhancing geometric factors in model learning and inference for object detection and instance segmentation[J]. IEEE Transactions on Cybernetics, 2021, 52(8): 8574-8586.

[29] TONG Z J, CHEN Y H, XU Z W, et al. Wise-IoU:Bounding box regression loss with dynamic focusing mechanism[J]. arXiv:2301.10051， 2023.

[30] MISRA D. Mish:A self regularized non-monotonic activation function[J]. arXiv:1908.08681， 2019.

基金资助

四川省科技创新苗子工程(2022040)