Few-shot cow identification via meta-learning

Cow identification is a prerequisite for precision livestock farming. Biometric-based methods have made significant progress in cow identification. However, substantial labelling costs and frequent identification task changes are still hamper model application. In this work, a novel method called “M...

Full description

Saved in:
Bibliographic Details
Main Authors: Xingshi Xu, Yunfei Wang, Yuying Shang, Guangyuan Yang, Zhixin Hua, Zheng Wang, Huaibo Song
Format: Article
Language:English
Published: Elsevier 2025-03-01
Series:Information Processing in Agriculture
Subjects:
Online Access:http://www.sciencedirect.com/science/article/pii/S2214317324000210
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850070797251510272
author Xingshi Xu
Yunfei Wang
Yuying Shang
Guangyuan Yang
Zhixin Hua
Zheng Wang
Huaibo Song
author_facet Xingshi Xu
Yunfei Wang
Yuying Shang
Guangyuan Yang
Zhixin Hua
Zheng Wang
Huaibo Song
author_sort Xingshi Xu
collection DOAJ
description Cow identification is a prerequisite for precision livestock farming. Biometric-based methods have made significant progress in cow identification. However, substantial labelling costs and frequent identification task changes are still hamper model application. In this work, a novel method called “MFCI” was proposed to achieve accurate cow identification under few-shot and task-changing conditions. Specifically, the proposed method comprises two components: cow location and cow identification. First, an improved YOLOv5n with Ghost module was adopted to quickly detect cow locations in images. Then, the Model-Agnostic Meta-Learning (MAML) framework was introduced for accurate identification under few-shot conditions and for fast adaptation to frequent changes in individual cows. Moreover, an autoencoder was adopted to allow Base-Learner learn more generalized features by combining both supervised and unsupervised approaches. The experimental results showed that the proposed cow location model achieved a mAP of 99.5 %. The proposed cow identification model attained an accuracy of 90.43 % with only five samples per cow for 20 cows, outperforming other state-of-the-art methods. The results demonstrate the broad applicability and significant value of the proposed method.
format Article
id doaj-art-e98b78d37a254ad79d4bcda550e8433b
institution DOAJ
issn 2214-3173
language English
publishDate 2025-03-01
publisher Elsevier
record_format Article
series Information Processing in Agriculture
spelling doaj-art-e98b78d37a254ad79d4bcda550e8433b2025-08-20T02:47:27ZengElsevierInformation Processing in Agriculture2214-31732025-03-01121809010.1016/j.inpa.2024.04.001Few-shot cow identification via meta-learningXingshi Xu0Yunfei Wang1Yuying Shang2Guangyuan Yang3Zhixin Hua4Zheng Wang5Huaibo Song6College of Mechanical and Electronic Engineering, Northwest A&F University, Yangling 712100, Shaanxi, China; Shaanxi Key Laboratory of Agricultural Information Perception and Intelligent Service, Yangling, 712100, Shaanxi, China; Key Laboratory of Agricultural Internet of Things, Ministry of Agriculture and Rural Affairs, Yangling 712100, Shaanxi, ChinaCollege of Mechanical and Electronic Engineering, Northwest A&F University, Yangling 712100, Shaanxi, China; Shaanxi Key Laboratory of Agricultural Information Perception and Intelligent Service, Yangling, 712100, Shaanxi, China; Key Laboratory of Agricultural Internet of Things, Ministry of Agriculture and Rural Affairs, Yangling 712100, Shaanxi, ChinaCollege of Mechanical and Electronic Engineering, Northwest A&F University, Yangling 712100, Shaanxi, China; Shaanxi Key Laboratory of Agricultural Information Perception and Intelligent Service, Yangling, 712100, Shaanxi, China; Key Laboratory of Agricultural Internet of Things, Ministry of Agriculture and Rural Affairs, Yangling 712100, Shaanxi, ChinaCollege of Mechanical and Electronic Engineering, Northwest A&F University, Yangling 712100, Shaanxi, China; Shaanxi Key Laboratory of Agricultural Information Perception and Intelligent Service, Yangling, 712100, Shaanxi, China; Key Laboratory of Agricultural Internet of Things, Ministry of Agriculture and Rural Affairs, Yangling 712100, Shaanxi, ChinaCollege of Mechanical and Electronic Engineering, Northwest A&F University, Yangling 712100, Shaanxi, China; Shaanxi Key Laboratory of Agricultural Information Perception and Intelligent Service, Yangling, 712100, Shaanxi, China; Key Laboratory of Agricultural Internet of Things, Ministry of Agriculture and Rural Affairs, Yangling 712100, Shaanxi, ChinaCollege of Mechanical and Electronic Engineering, Northwest A&F University, Yangling 712100, Shaanxi, China; Shaanxi Key Laboratory of Agricultural Information Perception and Intelligent Service, Yangling, 712100, Shaanxi, China; Key Laboratory of Agricultural Internet of Things, Ministry of Agriculture and Rural Affairs, Yangling 712100, Shaanxi, ChinaCorresponding author.; College of Mechanical and Electronic Engineering, Northwest A&F University, Yangling 712100, Shaanxi, China; Shaanxi Key Laboratory of Agricultural Information Perception and Intelligent Service, Yangling, 712100, Shaanxi, China; Key Laboratory of Agricultural Internet of Things, Ministry of Agriculture and Rural Affairs, Yangling 712100, Shaanxi, ChinaCow identification is a prerequisite for precision livestock farming. Biometric-based methods have made significant progress in cow identification. However, substantial labelling costs and frequent identification task changes are still hamper model application. In this work, a novel method called “MFCI” was proposed to achieve accurate cow identification under few-shot and task-changing conditions. Specifically, the proposed method comprises two components: cow location and cow identification. First, an improved YOLOv5n with Ghost module was adopted to quickly detect cow locations in images. Then, the Model-Agnostic Meta-Learning (MAML) framework was introduced for accurate identification under few-shot conditions and for fast adaptation to frequent changes in individual cows. Moreover, an autoencoder was adopted to allow Base-Learner learn more generalized features by combining both supervised and unsupervised approaches. The experimental results showed that the proposed cow location model achieved a mAP of 99.5 %. The proposed cow identification model attained an accuracy of 90.43 % with only five samples per cow for 20 cows, outperforming other state-of-the-art methods. The results demonstrate the broad applicability and significant value of the proposed method.http://www.sciencedirect.com/science/article/pii/S2214317324000210Precision livestockCattle identificationFew-shotTask-drivenAutoencoder
spellingShingle Xingshi Xu
Yunfei Wang
Yuying Shang
Guangyuan Yang
Zhixin Hua
Zheng Wang
Huaibo Song
Few-shot cow identification via meta-learning
Information Processing in Agriculture
Precision livestock
Cattle identification
Few-shot
Task-driven
Autoencoder
title Few-shot cow identification via meta-learning
title_full Few-shot cow identification via meta-learning
title_fullStr Few-shot cow identification via meta-learning
title_full_unstemmed Few-shot cow identification via meta-learning
title_short Few-shot cow identification via meta-learning
title_sort few shot cow identification via meta learning
topic Precision livestock
Cattle identification
Few-shot
Task-driven
Autoencoder
url http://www.sciencedirect.com/science/article/pii/S2214317324000210
work_keys_str_mv AT xingshixu fewshotcowidentificationviametalearning
AT yunfeiwang fewshotcowidentificationviametalearning
AT yuyingshang fewshotcowidentificationviametalearning
AT guangyuanyang fewshotcowidentificationviametalearning
AT zhixinhua fewshotcowidentificationviametalearning
AT zhengwang fewshotcowidentificationviametalearning
AT huaibosong fewshotcowidentificationviametalearning