PUFFER-DETR: Tiger puffer similar abnormal behavior recognition based on transformer

IF 3.6 2区农林科学 Q2 AGRICULTURAL ENGINEERING

Aquacultural Engineering Pub Date : 2025-04-26 DOI:10.1016/j.aquaeng.2025.102552

Yixi Zhang , Zeyuan Hu , Jihang Liu , Yinjia Li , Jianjian Lin , Yue Wang , Hong Yu

{"title":"PUFFER-DETR: Tiger puffer similar abnormal behavior recognition based on transformer","authors":"Yixi Zhang , Zeyuan Hu , Jihang Liu , Yinjia Li , Jianjian Lin , Yue Wang , Hong Yu","doi":"10.1016/j.aquaeng.2025.102552","DOIUrl":null,"url":null,"abstract":"<div><div>Fish behavior monitoring is crucial for fish farmers to obtain growth information, improve aquatic product quality, and adjust aquaculture strategies. However, the small size, severe occlusion, and similar behavior of fish pose challenges for identifying abnormal behavior. Therefore, this study proposes an abnormal behavior detection method based on PUFFER-DETR. Using the TripletAttention backbone network, the ability of the model to extract features of fish swarm behavior and individual fish behavior in turbid water has been improved. Furthermore, weight calculation is performed on the similar behavioral characteristics between individual fish and the behavioral characteristics of the fish groups to obtain a relationship feature map of similar behavior. Cross-scale feature fusion is performed using SHS-FPN, and the similarity behavior features of individual fish are adjusted to avoid the loss of similarity behavior features during the feature fusion process. Experimental results indicate that PUFFER-DETR achieved the best fusion accuracy at a speed of 127.9 frames per second, with an average accuracy of 92.8 %. This method can accurately detect abnormal behavior of fish and assist aquaculture personnel in judging the growth status of fish. Source code is available at <span><span>https://github.com/DLOU-FishBehavior/PUFFER-DETR</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":8120,"journal":{"name":"Aquacultural Engineering","volume":"111 ","pages":"Article 102552"},"PeriodicalIF":3.6000,"publicationDate":"2025-04-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Aquacultural Engineering","FirstCategoryId":"97","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S014486092500041X","RegionNum":2,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"AGRICULTURAL ENGINEERING","Score":null,"Total":0}

引用次数: 0

Abstract

Fish behavior monitoring is crucial for fish farmers to obtain growth information, improve aquatic product quality, and adjust aquaculture strategies. However, the small size, severe occlusion, and similar behavior of fish pose challenges for identifying abnormal behavior. Therefore, this study proposes an abnormal behavior detection method based on PUFFER-DETR. Using the TripletAttention backbone network, the ability of the model to extract features of fish swarm behavior and individual fish behavior in turbid water has been improved. Furthermore, weight calculation is performed on the similar behavioral characteristics between individual fish and the behavioral characteristics of the fish groups to obtain a relationship feature map of similar behavior. Cross-scale feature fusion is performed using SHS-FPN, and the similarity behavior features of individual fish are adjusted to avoid the loss of similarity behavior features during the feature fusion process. Experimental results indicate that PUFFER-DETR achieved the best fusion accuracy at a speed of 127.9 frames per second, with an average accuracy of 92.8 %. This method can accurately detect abnormal behavior of fish and assist aquaculture personnel in judging the growth status of fish. Source code is available at https://github.com/DLOU-FishBehavior/PUFFER-DETR.

查看原文本刊更多论文

puffer - detr：基于变压器的老虎puffer相似异常行为识别

鱼类行为监测对养殖户获取生长信息、提高水产品质量、调整养殖策略具有重要意义。然而，鱼类体型小，闭塞严重，行为相似，为识别异常行为带来了挑战。因此，本研究提出了一种基于PUFFER-DETR的异常行为检测方法。利用TripletAttention骨干网络，提高了模型对浑浊水中鱼群行为和个体行为特征的提取能力。在此基础上，对个体间的相似行为特征与鱼群间的相似行为特征进行权重计算，得到相似行为的关系特征图。采用SHS-FPN进行跨尺度特征融合，并对鱼类个体的相似行为特征进行调整，避免了特征融合过程中相似行为特征的丢失。实验结果表明，PUFFER-DETR在127.9帧/秒的速度下达到了最好的融合精度，平均精度为92.8 %。该方法可以准确检测鱼类的异常行为，辅助养殖人员判断鱼类的生长状况。源代码可从https://github.com/DLOU-FishBehavior/PUFFER-DETR获得。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文求助全文

来源期刊

Aquacultural Engineering 农林科学-农业工程

CiteScore

8.60

自引率

10.00%

发文量

审稿时长

>24 weeks

期刊介绍： Aquacultural Engineering is concerned with the design and development of effective aquacultural systems for marine and freshwater facilities. The journal aims to apply the knowledge gained from basic research which potentially can be translated into commercial operations. Problems of scale-up and application of research data involve many parameters, both physical and biological, making it difficult to anticipate the interaction between the unit processes and the cultured animals. Aquacultural Engineering aims to develop this bioengineering interface for aquaculture and welcomes contributions in the following areas: – Engineering and design of aquaculture facilities – Engineering-based research studies – Construction experience and techniques – In-service experience, commissioning, operation – Materials selection and their uses – Quantification of biological data and constraints