计算机科学 ›› 2024, Vol. 51 ›› Issue (6A): 230500176-7.doi: 10.11896/jsjkx.230500176
谌雨章, 王诗琦, 周雯, 周婉婷
CHEN Yuzhang, WANG Shiqi, ZHOU Wen, ZHOU Wanting
摘要： 为解决因水下成像环境退化导致图像分辨率较低,以及因鱼群目标较小等因素导致的检测精度不高的问题,提出了一种结合SPD-Conv结构和NAM注意力机制的改进YOLOv7检测算法。首先,采用Space-to-Depth(SPD)结构改进头部网络,取代了网络中原有的跨步卷积结构,保留了更多的细粒度信息,提升了特征学习的效率,提高了网络对低分辨率图像的检测效果。然后在网络中引入Normalization-based Attention Module(NAM)注意力机制,采用CBAM的模块集成方式,使用BN缩放因子来计算注意力权重,抑制了不显著的特征,提升了小目标检测的准确率。最后针对水下成像退化,对检测图片做反卷积预处理,减小了水下成像退化因素对检测造成的影响。实验结果显示,在WildFish数据集上模型的整体精度达到97.2%,与YOLOv7算法相比提升了7.6%,准确率提升了8.5%,召回率提升了9.8%,与Efficientdet,SSD,YOLOv5及YOLOv8算法相比,所提模型精度分别提升了12.6%,17.8%,4%及2.9%,在Aquarium数据集上模型的整体精度达到80.5%,相比Efficientdet,SSD,YOLOv5,YOLOv7及YOLOv8分别提升了18.4%,11.6%,6.9%,2.0%及2.7%,可以满足水下鱼群识别的需求。
中图分类号:
| [1]ID L,MIAO Z,PENG F,et al.Automatic counting methods in aquaculture:a review[J].Journal of the World Aquaculture Society,2021,52(2):269-283. [2]FAN L Z,LIU Y.Automate fry counting using computer vision and multi-class least squares support vector machine[J].Aquaculture,2013,380/381/382/383:91-98. [3]GIRSHICK R,DONAHUE J,DARRELL T,et al.Rich feature hierarchies for accurate object detection and semantic segmentation[C]//2014 IEEE Conference on Computer Vision and Pattern Recognition.Columbus,OH,USA.IEEE,2014:580-587. [4]LIU W,ANGUELOV D,ERHAND,et al.Ssd:Single shotmultibox detector[C]//European Conference on Computer Vision.Cham:Springer,2016:21-37. [5]REDMON J,DIVVALA S,GIRSHICK R,et al.You only look once:Unified,real-time object detection[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.2016:779-788. [6]REDMON J,FARHADIA.YOLO9000:better,faster,stronger[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.2017:7263-7271. [7]REDMON J,FARHADI A.Yolov3:An incremental improve-ment[J].arXiv:1804.02767,2018. [8]TSENG C H,KUO C H.Detecting and counting harvested fish and identifying fish types in electronic monitoring system videos using deep convolutional neural networks[J].ICES Journal of Marine Science,2020,77(4):1367-1378. [9]ZENG L C,SUN B,ZHU D Q.Underwater target detectionbased on Faster R-CNN and adversarial occlusion network[J].Engineering Applications of Artificial Intelligence,2021,100:104190. [10]ZHAO D,YANG B,DOU Y,et al.Underwater fish detection in sonar image based on an improved Faster RCNN[C]//2022 9th International Forum on Electrical Engineering and Automation(IFEEA).Zhuhai,China,2022:358-363. [11]SHEN J Y,LI L Y,DAI Y L,et al.A fish stock detection method based on feature fusion SSD[J].Computer Simulation,2020,37(11):422-426,469. [12]ZHANG L,HUANG L,LI B B,et al.Fish counting method based on multi-scale fusion and anchorless YOLO v3[J].Transactions of the Chinese Society for Agricultural Machinery,2021,52(S1):237-244. [13]ABDULLAH A M,FAKHRUL H,MD F H B E,et al.Fahad Hasan Bhuiyan EMON,et al.YOLO-Fish:A robust fish detection model to detect fish in realistic underwater environment[J].Ecological Informatics,2022,72:101847. [14]ZHAO S L,ZHANG S,LU J M,et al.A lightweight dead fish detection method based on deformable convolution and YOLOV4[J].Computers and Electronics in Agriculture,2022,198:107098. [15]ZHANG Y S,XU W X,YANG S S,et al.Improved YOLOX detection algorithm for contraband in X-ray images[J].Applied Optics,2022,61:6297-6310. [16]VIJIYAKUMAR K,GOVINDASAMY V,AKILA G.Hybridi-zation of Deep Convolutional Neural Network for Underwater Object Detection and Tracking Model[J].Microprocessors and Microsystems,2022,94:104628. [17]WANG C Y,BOCHKOVSKIY Al,LIAO H Y.YOLOv7:Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors[J].arXiv:2207.02696,2022. [18]SUNKARA R,LUO T.No More Strided Convolutions or Pooling:A New CNN Building Block for Low-Resolution Images and Small Objects[J].arXiv:2208.03641,2022. [19]LIU Y,SHAO Z,TENG Y,et al.NAM:Normalization-basedAttention Module[J].arXiv,abs/2111.12419,2021. [20]MIAO Y.Underwater image adaptive restoration and analysisby turbulence model[C]//2012 World Congress on Information and Communication Technologies.IEEE,2012:1182-1187. [21]ZHUANG P,WANG Y,QIAO Y Y.Wildfish:A large bench-mark for fish recognition in the wild[C]//Proceedings of the 26th ACM international conference on Multimedia.2018:1301-1309. [22]Roboflow.Aquarium Combined Image Dataset[EB/OL].https://roboflow.com |
|
||
首页