Application of Mask R-CNN Model in Identification of Eggplant Flowering Period

doi:10.3778/j.issn.1002-8331.2102-0241

Abstract

Abstract: The Mask R-CNN example segmentation model is applied to the identification of eggplant flowering stage. In order to solve the problems of misdetection and omission of large targets, a method of using hybrid dilated convolution to fuse ordinary convolution. The parameters are modified in the residual blocks of ResNet50. The whole feature extraction network is improved by stacking the residual blocks, the feature map perception field is expanded, and the global information correlation is enhanced. For the over-fitting problem, the pre-training ResNet50 feature extraction network is used as the initial parameter of the eggplant flower recognition model by using the transfer learning method, which improves the generalization ability of the test set and the training speed of the model. The mAP and mIOU of the improved model are 0.962 and 0.715 on the test set. By comparison with other models, the improved model can effectively improve the large target segmentation ability and has a good effect on eggplant florescence recognition. This study provides technical support for automatic pollination and florescence management of eggplant, which is important for ensuring pollination quality and improving economic benefits.

Key words: Mask R-CNN model, instance segmentation, target detection, hybrid dilated convolution, transfer learning

摘要： 将Mask R-CNN实例分割模型应用于茄子花期识别研究，针对Mask R-CNN模型对大目标物存在误检和漏检的情况，提出使用混合空洞卷积融合普通卷积的方法，在ResNet50的残差块中进行参数修改，通过堆叠残差块完成对整个特征提取网络的改进，扩大了特征图感受野，增强了全局信息关联性；针对出现的过拟合问题，运用迁移学习方法将预训练的ResNet50特征提取网络作为茄花识别模型的初始参数，提高了模型在测试集泛化能力的同时提升了模型训练速度。运用改进的模型在测试集上的mAP为0.962，mIOU为0.715。通过定性分析并与其他模型进行对比，证明改进的模型能有效提高大目标物分割能力，对茄子花期识别具有良好效果。该研究为茄花自动授粉与花期管理提供了技术支持，对保证授粉质量，提升经济效益具有重要意义。

关键词: Mask R-CNN模型, 实例分割, 目标检测, 混合空洞卷积, 迁移学习

ZHENG Kai, FANG Chun, YUAN Simiao, FENG Chuang, LI Guokun. Application of Mask R-CNN Model in Identification of Eggplant Flowering Period[J]. Computer Engineering and Applications, 2022, 58(18): 318-326.

郑凯, 方春, 袁思邈, 冯创, 李国坤. Mask R-CNN模型在茄花花期识别中的应用研究[J]. 计算机工程与应用, 2022, 58(18): 318-326.

References

[1] 刘晖，李兆雄，詹杰，等.机器视觉技术在农作物生长状况监测的研究进展[J].福建农机，2018，154（4）：22-28.
LIU H，LI Z X，ZHAN J，et al.Research progress of machine vision technology in crop growth monitoring[J].Fujian Agricultural Machinery，2018，154（4）：22-28.
[2] ANDREA C C，DANIEL B，MISAEL J.Precise weed and maize classification through convolutional neuronal networks[C]//2017 IEEE Second Ecuador Technical Chapters Meeting（ETCM），2017：1-6.
[3] 陈建能，张国凤.计算机视觉技术在农业中的应用及展望[J].甘肃农业大学学报，2003，38（2）：248-253.
CHEN J N，ZHANG G F.Application and prospect of computer vision technique in agriculture[J].Journal of Gansu Agricultural University，2003，38（2）：248-253.
[4] 王润涛，张长利，房俊龙，等.基于机器视觉的大豆籽粒精选技术[J].农业工程学报，2011，27（8）：355-359.
WANG R T，ZHANG C L，FANG J L，et al.Soybean seeds selection based on computer vision[J].Transactions of the Chinese Society of Agricultural Engineering，2011，27（8）：355-359.
[5] NIE P，ZHANG J，FENG X，et al.Classification of hybrid seeds using near-infrared hyperspectral imaging technology combined with deep learning[J].Sensors and Actuators B：Chemical，2019，296：126630.
[6] ESPEJO-GARCIA B，MYLONAS N，ATHANASAKOS L，et al.Combining generative adversarial networks and agricultural transfer learning for weeds identification[J].Biosystems Engineering，2021，204：79-89.
[7] FERENTINOS K P.Deep learning models for plant disease detection and diagnosis[J].Computers and Electronics in Agriculture，2018，145：311-318.
[8] 毕智健，张若宇，齐妍杰，等.基于机器视觉的番茄成熟度颜色判别[J].食品与机械，2016，32（12）：133-136.
BI Z J，ZHANG R Y，QI Y J，et al.Tomato maturity color discrimination based on machine vision[J].Food and Machinery，2016，32（12）：133-136.
[9] HAN Z Z，LIU J，ZHAO Y G，et al.Grading system of pear’s appearance quality based on computer vision[C]//International Conference on Systems and Informatics，2012.
[10] FENG J，WANG Z W，ZHA M，et al.Flower recognition based on transfer learning and adamdeep learning optimization algorithm[C]//The 2019 International Conference，2019：598-604.
[11] 岳有军，李想，赵辉，等.一种基于计算机视觉的花蕾判别方法：CN104318240B[P].2017.
YUE Y J，LI X，ZHAO H，et al.A flower bud discrimination method based on computer vision：CN104318240B[P].
2017.
[12] TIAN Y，YANG G，WANG Z，et al.Instance segmentation of apple flowers using the improved mask R-CNN model[J].Biosystems Engineering，2020，193：264-278.
[13] 刘艳平.柑橘采摘机器人障碍物识别与定位方法研究[D].重庆：重庆理工大学，2019.
LIU Y P.Research on obstacle recognition and location method of citrus picking robot[D].Chongqing：Chongqing University of Technology，2019.
[14] HE K M，GKIOXARI G，DOLLAR P.Mask R-CNN[J].arXiv：1703.06870，2017.
[15] WANG P，CHEN P，YUAN Y，et al.Understanding convolution for semantic segmentation[C]//2018 IEEE Winter Conference on Applications of Computer Vision（WACV），2018：1451-1460.
[16] COUTEAUX V，SI-MOHAMED S，NEMPONT O，et al.Automatic knee meniscus tear detection and orientation classification with Mask-RCNN[J].Diagnostic and Interventional Imaging，2019，100（4）：235-242.
[17] YU F，KOLTUN V，FUNKHOUSER T.Dilated residual networks[J].arXiv：1705.09914，2017.
[18] GIRSHICK R.Fast R-CNN[J].arXiv：1504.08083，2015.
[19] 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 & Machine Intelligence，2017，39（6）：1137-1149.
[20] PINHEIRO P O，COLLOBERT R，DOLLAR P.et al.Learning to segment object candidates[J].arXiv：1506.06204，2015.
[21] BADRINARAYANAN V，KENDALL A，CIPOLLA R.SegNet：a deep convolutional encoder-decoder architecture for image segmentation[J].IEEE Transactions on Pattern Analysis and Machine Intelligence，2017，39（12）：2481-2495.
[22] PAN S J，YANG Q.A survey on transfer learning[J].IEEE Transactions on Knowledge & Data Engineering，2010，22（10）：1345-1359.
[23] 赖佩霞，王晓东，章联军.基于迁移学习的蔬菜图像识别方法[J].宁波大学学报（理工版），2019，32（5）：36-41.
LAI P X，WANG X D，ZHANG L J.Vegetable image recognition method based on transfer learning[J].Journal of Ningbo University（Natural Science and Engineering Edition），2019，32（5）：36-41.