An ultra-lightweight method for individual identification of cow-back pattern images in an open image set
文献类型: 外文期刊
第一作者: Wang, Rong
作者: Wang, Rong;Gao, Ronghua;Li, Qifeng;Zhao, Chunjiang;Ding, Luyu;Yu, Ligen;Ma, Weihong;Wang, Rong;Zhao, Chunjiang;Gao, Ronghua;Li, Qifeng;Zhao, Chunjiang;Ding, Luyu;Yu, Ligen;Ma, Weihong;Ru, Lin
作者机构: Beijing Acad Agr & Forestry Sci, Informat Technol Res Ctr, Beijing 100097, Peoples R China;Northwest A&F Univ, Coll Informat Engn, Yangling 712100, Peoples R China;Natl Engn Res Ctr Informat Technol Agr, Beijing 100097, Peoples R China;Heilongjiang Bayi Agr Univ, Coll Civil Engn & Water Conservancy, Daqing 163319, Peoples R China
关键词: Cow-back pattern; Cow recognition; LightCowsNet; Open image set; Deep learning
期刊名称:EXPERT SYSTEMS WITH APPLICATIONS ( 2022影响因子:8.5; 五年影响因子:8.3 )
ISSN: 0957-4174
年卷期: 2024 年 249 卷
收录情况: SCI
摘要: Cow recognition forms the foundation of smart livestock management. Current closed-set cow recognition models can exclusively identify cows within their training set and are ill-equipped to recognize other subjects in the open world. Additionally, there is a dearth of comprehensive research on open-set cow recognition. Existing open-set recognition models not only exhibit suboptimal accuracy but also impose a substantial burden on hardware resources due to their extensive parameterization. These challenges hinder the practical deployment of cow recognition models. This study introduces an innovative open-set metric learning-based recognition framework for cow-back patterns, aiming to rectify a significant limitation of individual cow recognition models, which is their inability to identify cows not included in the training set. The proposed framework combines various loss functions, metric methods, and backbone networks. This integration enables the open-set recognition of cow back pattern images, thereby facilitating the identification of cows previously unrecognized by the model. To mitigate hardware resource consumption, we have devised a novel ultra-lightweight backbone network known as LightCowsNet. This network draws inspiration from existing lightweight backbone networks such as MobileFaceNet, MobileViT, and EfficientNetV2. Its purpose is to extract features from cow back pattern images efficiently. LightCowsNet employs attention mechanisms, inverted residual structures, and depthwise separable convolutions. LightCowsNet first uses attention mechanisms, inverted residual structures, and depthwise separable convolutions to design multiple new feature extraction modules. These modules, which have different structures, are subsequently placed to extract different levels of image information, which enables complete extraction of the textural information of the cows' back patterns. The present study reorganized the Cows2021 dataset to render it suitable for open-set recognition of cow-back patterns, with the test set comprising image pairs. The results revealed that using A-SoftMax as the loss function and Euclidean distance as the metric method enabled LightCowsNet to achieve the highest accuracy of 94.26 %, with a model weight space occupation of only 4.06 MB. Compared to MobileFaceNet, MobileViT, and EfficientNetV2, LightCowsNet achieved 6.24 %, 8.82 %, and 2.3 % greater accuracy on the test set, respectively. Its weight also decreased by 0.71 MB, 0.24 MB, and 4.3 MB, respectively. These results demonstrate that the proposed model achieved both accuracy improvement and model lightweightness, and these findings could serve as reference solutions for intelligent farming on cow farms.
- 相关文献
作者其他论文 更多>>
-
A Clean and Health-Care-Focused Way to Reduce Indoor Airborne Bacteria in Calf House with Long-Wave Ultraviolet
作者:Ding, Luyu;Yao, Chunxia;Li, Qifeng;Ding, Luyu;Yao, Chunxia;Li, Qifeng;Ding, Luyu;Yao, Chunxia;Li, Qifeng;Zhang, Qing;Wang, Chaoyuan;Shan, Feifei
关键词:closed calf house; emission rate; size distribution; microbial composition; health improvement
-
An FPGA implementation of Bayesian inference with spiking neural networks
作者:Li, Haoran;An, Lingling;Wan, Bo;An, Lingling;Wan, Bo;Fang, Ying;Fang, Ying;Li, Qifeng;Liu, Jian K.
关键词:spiking neural networks; probabilistic graphical models; Bayesian inference; importance sampling; FPGA
-
Recognition of wheat rusts in a field environment based on improved DenseNet
作者:Chang, Shenglong;Cheng, Jinpeng;Fan, Zehua;Ma, Xinming;Li, Yong;Zhao, Chunjiang;Chang, Shenglong;Yang, Guijun;Cheng, Jinpeng;Fan, Zehua;Yang, Xiaodong;Zhao, Chunjiang
关键词:Plant disease; Wheat rust; Image processing; Deep learning; Computer vision (CV); DenseNet
-
GCVC: Graph Convolution Vector Distribution Calibration for Fish Group Activity Recognition
作者:Zhao, Zhenxi;Zhao, Chunjiang;Zhao, Zhenxi;Yang, Xinting;Zhou, Chao;Zhao, Chunjiang;Zhao, Zhenxi;Yang, Xinting;Zhou, Chao;Zhao, Chunjiang;Zhao, Zhenxi;Yang, Xinting;Zhou, Chao;Zhao, Chunjiang;Liu, Jintao
关键词:Fish; Feature extraction; Activity recognition; Calibration; Adhesives; Training; Convolution; Graph convolution vector calibration; fish group activity; activity feature vector calibration; fish activity dataset
-
Adaptive precision cutting method for rootstock grafting of melons: modeling, analysis, and validation
作者:Chen, Shan;Zhao, Chunjiang;Chen, Shan;Jiang, Kai;Zheng, Wengang;Jia, Dongdong;Zhao, Chunjiang;Jiang, Kai;Zheng, Wengang;Jia, Dongdong;Zhao, Chunjiang
关键词:Melon; Grafting robot; Adaptive cutting; Rootstock pith cavity; Machine vision
-
Long-range infrared absorption spectroscopy and fast mass spectrometry for rapid online measurements of volatile organic compounds from black tea fermentation
作者:Yang, Chongshan;Li, Guanglin;Zhao, Chunjiang;Fu, Xinglan;Yang, Chongshan;Jiao, Leizi;Wen, Xuelin;Lin, Peng;Duan, Dandan;Zhao, Chunjiang;Dong, Daming;Yang, Chongshan;Jiao, Leizi;Wen, Xuelin;Lin, Peng;Duan, Dandan;Dong, Daming;Dong, Chunwang
关键词:Black tea fermentation; Volatile organic compounds; Proton transfer reaction mass spectrometry; Fourier transform infrared spectroscopy; Principal component analysis; Extreme learning machine
-
A Point Cloud Segmentation Method for Pigs from Complex Point Cloud Environments Based on the Improved PointNet++
作者:Chang, Kaixuan;Xu, Xingmei;Li, Qifeng;Ma, Weihong;Xue, Xianglong;Xu, Zhankang;Li, Mingyu;Guo, Yuhang;Meng, Rui;Li, Qifeng;Ma, Weihong;Qi, Xiangyu;Xue, Xianglong;Li, Mingyu;Guo, Yuhang;Meng, Rui;Li, Qifeng;Ma, Weihong;Xue, Xianglong;Li, Mingyu;Guo, Yuhang;Meng, Rui;Li, Qifeng;Xu, Zhankang
关键词:point cloud segmentation; PointNet++; 3D point cloud processing; SoftPool