基于轻量化YOLO⁃v3的绿熟期番茄检测方法

doi:10.13733/j.jcam.issn.2095⁃5553.2022.03.018

摘要/Abstract

摘要： 准确识别定位绿熟期番茄果实是实现其自动采摘的必要前提。由于绿熟期番茄的表面颜色仍为青色与叶片、枝干颜色接近，特别是存在叶片、枝干遮挡和果实重叠类型的图像，传统的图像检测处理方法不能准确进行定位。为解决此问题，采用改进的深度学习目标检测算法YOLO⁃v3进行番茄检测，将原算法的骨干网络DarkNet⁃53改为更轻量化的Mobilenet⁃v1。结果表明：轻量化YOLO⁃v3算法将模型大小缩小为原来的39.38%，训练速度提高3.88倍，验证集的平均精度均值达到98.69%，测试集的平均精度均值达到98.28%。所采用的轻量化YOLO⁃v3检测算法可实现对绿熟期番茄的实时目标检测，更适合在移动设备和嵌入式端进行部署，为更加高效的番茄自动采摘奠定基础。

关键词: 绿熟期番茄, 自动采摘, 轻量化, YOLO?v3算法

Abstract: The accurate detection of mature green tomato is a prerequisite of automatic picking. Owing to color similarity between the surface color of mature green tomato, leaf, and branch, as well as the existence of images with leaf, branch occlusion and fruit overlapping, the detection performance of traditional image detection was bad. To solve this problem, an improved YOLO⁃v3 algorithm was designed for mature green tomato image detection, where the original DarkNet⁃53 backbone was replaced by the lightweight Mobilenet⁃v1. Compared with the original YOLO⁃v3 algorithm, the lightweight algorithm reduced the model size by 39.38%, increased the train speed by 3.88 times, and the mean average precision of the validation set and test set was 98.69% and 98.28%, respectively. Therefore, the proposed lightweight YOLO⁃v3 could successfully realize real⁃time detection of mature green tomato and is more suitable to be implement in mobile and embedded devices, which will make automatic picking of tomato more efficient.

Key words: mature green tomato, automatic picking, lightweight, YOLO?v3 algorithm

中图分类号:

S641.2： TP391.4

苏斐, 张泽旭, 赵妍平, 李天华, 祖林禄, . 基于轻量化YOLO⁃v3的绿熟期番茄检测方法[J]. 中国农机化学报, 2022, 43(3): 132-137.

Su Fei, Zhang Zexu, Zhao Yanping, Li Tianhua, Zu Linlu. . Detection of mature green tomato based on lightweight YOLO⁃v3[J]. Journal of Chinese Agricultural Mechanization, 2022, 43(3): 132-137.

参考文献

[1] Kapach K, Barnea E, Mairon R, et al. Computer vision for fruit harvesting robots⁃state of the art and challenges ahead [J]. International Journal of Computational Vision & Robotics, 2012, 3(1/2): 4-34.
[2] 周俊, 刘锐, 张高阳. 基于立体视觉的水果采摘机器人系统设计[J]. 农业机械学报, 2010, 41(6): 158-162.
Zhou Jun, Liu Rui, Zhang Gaoyang. Design of fruit picking robot system based on stereo vision [J]. Transactions of the Chinese Society for Agricultural Machinery, 2010, 41(6): 158-162.
[3] 阮承治, 赵德安, 陈旭, 等. 双指型农业机器人抓取球形果蔬的控制器设计[J]. 中国农机化学报, 2019, 40(11): 169-175.
Ruan Chengzhi, Zhao Dean, Chen Xu, et al. Controller design for realizing double⁃finger agricultural robot to grasp spherical fruits and vegetables [J]. Journal of Chinese agricultural mechanization, 2019, 40(11): 169-175.
[4] 李寒, 张漫, 高宇, 等. 温室绿熟番茄机器视觉检测方法[J]. 农业工程学报, 2017, 33(Z1): 328-334, 388.
Li Han, Zhang Man, Gao Yu, et al. Green ripe tomato detection method based on machine vision in greenhouse [J]. Transactions of the Chinese Society of Agricultural Engineering, 2017, 33(Z1): 328-334, 388.
[5] Yamamoto K, Guo W, Yoshioka Y, et al. On plant detection of intact tomato fruits using image analysis and machine learning methods [J]. Sensors, 2014, 14(7): 12191-12206.
[6] 赵杰文, 刘木华, 杨国彬. 基于HIS颜色特征的田间成熟番茄识别技术[J]. 农业机械学报, 2004, 35(5): 122-124, 135.
Zhao Jiewen, Liu Muhua, Yang Guobin. Discrimination of mature tomato based on HIS color space in natural outdoor scenes [J]. Transactions of the Chinese Society for Agricultural Machinery, 2004, 35(5): 122-124, 135.
[7] 陶彦辉, 尹君驰, 方菲. 基于RGB颜色模型的番茄识别系统设计[J]. 机械研究与应用, 2015, 28(1): 159-160, 163.
Tao Yanhui, Yin Junchi, Fang Fei. Design of tomato identification system based on RGB color model [J]. Mechanical Research and Application, 2015, 28(1): 159-160, 163.
[8] 王红珠, 崔永杰. 基于计算机视觉的番茄颜色检测研究[J]. 农业技术与装备, 2017(1): 49-51.
Wang Hongzhu, Cui Yongjie. Tomato color detection research based on computer vision [J]. Agricultural Technology and Equipment, 2017(1): 49-51.
[9] 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.
Ma Cuihua, Zhang Xueping, Li Yutao, et al. Identification of immature tomatoes base on salient region detection and improved Hough transform method [J]. Transactions of the Chinese Society of Agricultural Engineering, 2016, 32(14): 219-226.
[10] Girshick R, Donahue J, Darrell T, et al. Rich feature hierarchies for accurate object detection and semantic segmentation [C]. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2014.
[11] Girshick R. Fast R⁃CNN [C]. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2015.
[12] Ren S, He K, Girshick R, et al. Faster R⁃CNN: Towards real⁃time object detection with region proposal networks [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2016, 39(6): 1137-1149.
[13] Liu W, Anguelov D, Erhan D, et al. SSD: Single shot multibox detector [C]. European Conference on Computer Vision, 2016.
[14] Redmon J, Farhadi A. YOLO9000: Better, faster, stronger [C]. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2017.
[15] 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.
[16] Redmon J, Farhadi A. YOLOv3: An incremental improvement [J]. arXiv preprint arXiv, 1804. 02767, 2018.
[17] 邢艳芳, 卓文鑫, 段红秀. 基于MobileNet的敏感图像识别系统设计[J]. 电视技术, 2018, 42(7): 53-56.
Xing Yanfang, Zhuo Wenxin, Duan Hongxiu. Design of sensitive image recognition system based on MobileNet [J]. Television Technology, 2018, 42(7): 53-56.
[18] 罗莎莎. 基于MobileNet⁃SSD模型的道路目标识别研究及其安卓应用开发[D]. 广州: 华南理工大学, 2019.
Luo Shasha. Road target recognition based on MobileNet⁃SSD model technical research and its Android APP development [D]. Guangzhou: South China University of Technology, 2019.
[19] Mao Q C, Sun H M, Liu Y B, et al. Mini⁃YOLOv3: Real⁃time object detector for embedded applications [J]. IEEE Access, 2019, 7: 133529-133538.
[20] Howard A G, Zhu Menglong, Chen Bo, et al. MobileNets: Efficient convolutional neural networks for mobile vision applications [J]. Computer Vision and Pattern Recognition arXiv, Preprint arXiv: 1704.04861, 2017.