基于CMYolov8的单环刺螠洞口检测
尹骏杨1 闵莹¹ 尹延峰2纪元1通讯作者

1中国农业大学烟台研究院，山东烟台，264670；

2马庄镇农业综合服务中心，山东泰安，271000；

摘要：本研究针对单环刺螠自动采捕和产量预测需求，提出了一种基于深度学习的单环刺螠洞口识别方法，并设计了适用于自动采捕船嵌入式设备的架构。该方法提出了一种新颖的CSPHet架构，取代了YOLOv8中的C2f结构。CSPHet通过HetConv实现异构核卷积，显著减少计算量和参数数量，同时保持高效的特征表示能力。模型通过1x1卷积预处理，分支网络分别保留原始特征和进行HetConv特征提取，并通过CSPHet_Bottleneck模块进一步深化特征表达。此外，模型融合了FPN-PAN结构及MetaNeXt模块，通过将大核卷积设计为Inception分支，增强位置信息的整合能力，提升了检测精度并减轻硬件部署负担。实验结果表明，改进后的网络在参数量和计算量上分别减少了16.7%和18.8%的同时，map50和map95上分别上涨了6.7%，13.9%。为实际应用场景下模型部署提供了模型基础。

关键词：单环刺螠洞口；目标检测；深度学习；YOLO v8

参考文献

[1]曹梅,辛益,陈吉圣,王兴强,梁笛,韦寿永,孙美超,徐明,于润川.单环刺螠特征特性及养殖技术[J].现代农业科技,2021,No.802(20):178-180.

[2]大连海洋大学.一种海肠采捕船：CN111528192A[P].2020-08-14.

[3]Wang, C.-Y.; Mark Liao, H.-Y.; Wu, Y.-H.; Chen, P.-Y.; Hsieh, J.-W.; Yeh, I-Hau. CSPNet: A New Backbone That Can Enhance Learning Capability of CNN. 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) 2020. https://doi.org/10.1109/cvprw50498.2020.00203.

[4]W. Yu, P. Zhou, S. Yan and X. Wang, "InceptionNeXt: When Inception Meets ConvNeXt," 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognitio n (CVPR), Seattle, WA, USA, 2024, pp. 5672-5683, doi: 10.1109/CVPR52733.2024.00542. keywords: {Performance evaluation;Training;Degradation;Costs;Computational modeling;Memory management;Network architecture;CNN;convolution;efficient neural networks},

[5] Liu, S., & Wang, Z. (2020). "CSPNet: A New Backbone that Can Enhance Learning Capability of CNN." Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 3907-3915. DOI: 10.1109/CVPR42600.2020.00397.

[6] Singh, P.; Vinay Kumar Verma; Rai, P.; Namboodiri, V. P. HetConv: Heterogeneous Kernel-Based Convolutions for Deep CNNs. Pure (University of Bath) 2019. https://doi.org/10.1109/cvpr.2019.00497.

[7] Lin, T. Y., Dollár, P., Girshick, R., He, K., Hariharan, B., & Belongie, S. (2017). "Feature Pyramid Networks for Object Detection." Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2117-2125. DOI: 10.1109/CVPR.2017.106.

[8] Xie, S., & Girshick, R. (2017). "Exploring Spatial Pyramid Pooling for Object Detection." IEEE Transactions on Pattern Analysis and Machine Intelligence, 39(4), 820-834. DOI: 10.1109/TPAMI.2016.2552714

[9] Tan, M., & Le, Q. V. (2020). "EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks." Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 6105-6114. DOI: 10.1109/CVPR42600.2020.00615