CA-YOLOv5: Detection model for healthy and diseased silkworms in mixed conditions based on improved YOLOv5

Hongkang Shi, Wenfu Xiao, Shiping Zhu, Linbo Li, Jianfei Zhang

Abstract


The accurate identification and localization of diseased silkworms is an important task in the research of disease precision control technology and equipment development in the sericulture industry. However, the existing deep learning-based methods for this task are mainly based on image classification, which fails to provide the location information of diseased silkworms. To this end, this study proposed an object detection-based method for identifying and locating healthy and diseased silkworms. Images of mixed healthy and diseased silkworms were collected using a mobile phone, and the category and location of each silkworm were labeled using LabelImg as a labeling tool to construct an image dataset for object detection. Based on the one-step detection model YOLOv5s, the ConvNeXt-Attention-YOLOv5 (CA-YOLOv5) model was designed in which the large kernel with depth-wise separable convolution (7×7 dw-conv) of ConvNeXt was adopted to expand receptive fields and the channel attention mechanism ECANet was added to enhance the capability of feature extraction. Experiments showed that the mean average precision (mAP) values of CA-YOLOv5 for healthy and diseased silkworms reached 96.46%, which is 1.35% better than that achieved via YOLOv5s. At the same time, the overall performance of CA-YOLOv5 was significantly better than state-of-the-art one-step models, such as Single Shot MultiBox Detector (SSD), CenterNet, and EfficientDet, and even improved YOLOv5 using image attention mechanism and a lightweight backbone, like SENet-YOLOv5 and MobileNet-YOLOv5. The results of this study can provide an important basis for the accurate positioning of diseased silkworms in precision disease control technology and equipment development.
Keywords: diseased silkworm detection, YOLOv5; mixed conditions, image attention mechanism, object detection
DOI: 10.25165/j.ijabe.20231606.7854

Citation: Shi H K, Xiao W F, Zhu S P, Li L B, Zhang J F. CA-YOLOv5: Detection model for healthy and diseased silkworms in mixed conditions based on improved YOLOv5. Int J Agric & Biol Eng, 2023; 16(6): 236–245.

Keywords


diseased silkworm detection, YOLOv5; mixed conditions, image attention mechanism, object detection

Full Text:

PDF

References


Jiang L, Zhao P, Xia Q Y. Research progress and prospect of silkworm molecular breeding for disease resistance. Acta Sericologica Sinica, 2014; 40（4）: 571–575. （in Chinese）

Xu A Y, Qian H Y, Sun P J, Liu M Z, Lin C L, Li G, et al. Breeding of a new silkworm variety Huakang 3 with resistance to Bombyx mori Nucleopolyhedrosis. Acta Sericologica Sinica, 2019; 45（2）: 201–211. （in Chinese）

Shi H K, Jiang M, Li L B, Wu J M, Ye J J, Ma Y, et al. Design of young silkworm feeding machine with spiral lifting system and its production test. Acta Sericologica Sinica, 2018; 44（6）: 891–897. （in Chinese）

Shi H K, Tian Y Y, Yang C, Chen Y, Su S Y, Zhang Z Y, et al. Research on intelligent recognition of silkworm larvae species based on convolutional neural network. Journal of Southwestern University （Natural Science Edition）, 2020; 42（12）: 34–45. （in Chinese）

Yu Y D, Gao P F, Zhao Y Z, Pan G Q, Chen T. Automatic male and female identification of silkworm pupae based on deep convolutional neural network. Acta Sericologica Sinica, 2020; 46（2）: 197–203. （in Chinese）

Li S J, Sun W H, Liang M, Shao T F, Shen J. Design of real-time cocoon species detection system based on deep learning. Shanghai Textile Science & Technology, 2021; 49（11）: 53–55,58. （in Chinese）

He R M, Zheng K F, Wei Q Y, Zhang X B, Zhang J, Zhu Y H, et al. Identification and counting of silkworms in factory farm using improved Mask R-CNN model. Smart Agriculture, 2022; 4（2）: 163–173. （in Chinese）

Wen C M, Wen J, Li J H, Luo Y Y, Chen M B, Xiao Z P, et al. Lightweight silkworm recognition based on Multi-scale feature fusion. Computers and Electronics in Agriculture, 2022; 200: 107234.

Shi H K, Xiao W F, Huang L, Hu C W, Hu G R, Zhang J F. Research on recognition of silkworm diseases based on Convolutional Neural Network. Journal of Chinese Agricultural Mechanization, 2022; 43（1）: 150–157. （in Chinese）

Xia D Y, Zhen Y, Cheng A J. Development and application of silkworm disease recognition system based on mobile App.Beijing: In: 10th International Conference on Image and Graphics （ICIG 2019）, Cham: Springer, 2019; pp.471-482.

Ding J Y, Cheng A J. An improved similarity algorithm based on deep hash and code bit independence. In: 2019 4th International Conference on Insulating Materials, Material Application and Electrical Engineering. 2019; 440: 032079.

Nicolas C, Francisco M, Gabriel S, Nicolas U, Alexander K, Sergey Z. End-to-end object detection with transformers. arXiv e-Prints archive, 2020. arXiv: 2005.12872.

Redmon J, Farhadi A. YOLOv3: An Incremental Improvement. arXiv e-Prints archive, 2018. arXiv: 1804.02767.

Glenn J. yolov5. Git code. Available: https://github.com/ultralytics/yolov5. Accessed on [2022-03-20

Wang C Y, Liao M H Y, Yeh I, Wu Y H, Chen P Y, Hsieh J W. CSPNet: A new backbone that can enhance learning capability of CNN. arXiv e-Prints archive, 2019. arXiv: 1911.11929.

Qi J T, Liu X N, Liu K, Xu F R, Guo H, Tian X L, et al. An improved YOLOv5 model based on visual attention mechanism: Application to recognition of tomato virus disease. Computer and Electronics in Agriculture, 2022; 194: 106780.

Ning J F, Lin J Y, Yang S Q, Wang Y S, Lan X Y. Face recognition method of dairy goat based on improved YOLO v5s. Transactions of the CSAM, 2023; 54（4）: 331–337. （in Chinese）

Yang Q M, Chen S B, Huang Y G, Xiao D Q, Liu Y F, Zhou J X. Pig counting algorithm based on improved YOLO v5n. Transactions of the CSAM, 2023; 54（1）: 251–262. （in Chinese）

Xue H X, Shen M X, Liu L S, Chen J X, Shan W P, Sun Y W. Estrus detection method of parturient sows based on improved YOLO v5s. Transactions of the CSAM, 2023; 54（1）: 263–270. （in Chinese）

Ma H X, Zhang M, Dong K B, Wei S H, Zhang R, Wang S X. Research of locust recognition in Ningxia grassland based on improved YOLO v5. Transactions of the CSAM, 2022; 53（11）: 270–279. （in Chinese）

Wang Q F, Cheng M, Huang S, Cai Z J, Zhang J L, Yuan H B. A deep learning approach incorporating YOLO v5 and attention mechanisms for field real-time detection of the invasive weed Solanum rostratum Dunal seedlings. Computers and Electronics in Agriculture, 2022; 199: 107194.

Li R, Wu Y P. Improved YOLO v5 wheat ear detection algorithm based on attention mechanism. Electronics, 2022; 11（11）: 1673.

Liu Z, Mao H, Wu C Y, Feichtenhofer C, Darrell T, Xie S. A ConvNet for the 2020s. 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022; pp.11966–11976.

Wang Q L, Wu B G, Zhu P F, Li P H, Zuo W M, Hu Q H. ECA-Net: Efficient channel attention for deep convolutional neural networks. In: 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition （CVPR）, IEEE, 2020; 11531–11539.

Qin Q, Dong Z Q, Lei X J, Cao M Y, Tang L, Shi M N, et al. Characteristic analysis of resistance to BmNPV in CVDAR strain of silkworm. Acta Microbiologica Sinica, 2019; 59（12）: 2390–2400. （in Chinese）

Dong Z Q, Lei X J, Qin Q, Zhang X L, Tang L, Shi M N, et al. Mechanism analysis of Anti-BmNPV resistant strain NC99R. Chinese Journal of Biotechnology, 2020; 36（1）: 100–108. （in Chinese）

Tzutalin D. LabelImg. Git code. 2015. Available:https://github.com/tzutalin/labelImg. Accessed on [2021-09-16

Zhang Z Z, Lan C L, Zeng W J, Jin X, Chen Z. Relation-aware global attention. arXiv e-prints archive, 2022. arXiv: 2201.03545.

Hu J, Shen L, Albanie S, Sun G, Wu E H. Squeeze-and-Excitation Networks. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2020; 42（8）: 2011–2023.

Loshchilov I, Hutter F. SGDR: Stochastic Gradient Descent with warm restarts. arXiv e-prints archive, 2017. arXiv: 1608.03983.

Rezatofighi H, Tsoi N, Gwak J Y, Sadeghian A, Reid I, Savarese S. Generalized intersection over union: A metric and a loss for bounding box regression. In: P2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Long Beach: IEEE, 2019; pp.658–666.

Bochkovskiy A, Wang C, Liao H M. YOLOv4: Optimal speed and accuracy of object detection. arXiv e-prints archive, 2020. arXiv: 2004.10934.

Zhou X, Wang D, Krähenbühl P. Objects as Points. arXiv e-prints archive, 2019. arXiv: 1904.07850.

Tan M X, Pang R M, Le Q V. EfficientDet: Scalable and efficient object detection. In: 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition （CVPR）, Seattle: IEEE, 2020; pp.10778–10787.

Liu W, Anguelov D, Erhan D, Szegedy C, Reed S, Fu C Y, et al. SSD: Single Shot MultiBox Detector. arXiv preprint arXiv: 1512.02325.

Ge Z, Liu S, Wang F, Li Z, Sun J. YOLOX: Exceeding YOLO series in 2021. arXiv e-prints archive, 2021. arXiv 2107.08430.

Cao Y Y, Chen J, Zhang Z C. A sheep dynamic counting scheme based on the fusion between an improved-sparrow-search YOLOv5x-ECA model and few-shot deepsort algorithm. Computers and Electronics in Agriculture, 2023; 206: 107696.

Lu S L, Chen W K, Zhang X, Karkee M. Canopy-attention-YOLOv4-based immature/mature apple fruit detection on dense-foliage tree architectures for early crop load estimation. Computer and Electronics in Agriculture, 2022; 193: 106696.

Xu L J, Wang Y H, Shi X S, Tang Z L, Chen X Y, Wang Y C, et al. Real-time and accurate detection of citrus in complex scenes based on HPL-YOLOv4. Computers and Electronics in Agriculture, 2023; 205: 107590.

Zhang L, Zhou X H, Li B B, Zhang H X, Duan Q L. Automatic shrimp counting method using local images and lightweight YOLOv4. Biosystems Engineering, 2022; 220: 39–54.




Copyright (c) 2023 International Journal of Agricultural and Biological Engineering

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 International License.

2023-2026 Copyright IJABE Editing and Publishing Office