Jiaxuan Yu , Guangxu Wang , Xin Li , Zhuangzhuang Du , Wenkai Xu , Muhammad Akhter , Daoliang Li
{"title":"SDYOLO-Tracker:一种高效的多鱼缺氧行为识别和跟踪方法","authors":"Jiaxuan Yu , Guangxu Wang , Xin Li , Zhuangzhuang Du , Wenkai Xu , Muhammad Akhter , Daoliang Li","doi":"10.1016/j.compag.2025.110079","DOIUrl":null,"url":null,"abstract":"<div><div>Real-time monitoring and tracking of fish in aquaculture and timely detection of fish hypoxia behavior play a crucial role in advancing the intelligent aquaculture. This study proposes an effective multi-fish behavior identification and tracking method (SDYOLO-Tracker) to solve the lag problem of hypoxia behavior monitoring in different species of fish. Our work aims to enable early monitoring of abnormal hypoxic behavior in fish. To enhance the model ability to detect small objects, we improved the YOLOv8n model using SPD-Conv and D-LKA modules. Then, we integrate the improved YOLOv8n (SDYOLOv8) model with Bytetrack multi-object tracking (MOT) algorithm. This approach utilizes a complementary 3D motion prediction strategy, which effectively addresses issues such as sudden fish motion and disappearance, resulting in an accurate multi-fish identification and tracking. The experimental results indicate that the enhanced MOTA, HOTA, IDR and IDF1 have shown improvement of 3.32 %, 3.53 %, 6.68 %, and 5.36 %, respectively when compared with the original YOLOv8n model. Moreover, in the tested video, the ID number switching is reduced by 18.75 %, significantly improving the accuracy of multi-fish tracking without compromising the speed of the model. In the comparative experiments with other MOT algorithms, this method achieved the highest IDR and IDF1 metrics, with a frame per second (FPS) of 45.03, demonstrating the best tracking stability and the fastest model processing speed. In addition, SDYOLO-Tracker can conduct qualitative and quantitative analyses of fish behavior under varying dissolved oxygen concentrations, reflecting changes in locomotion, average velocity, and maximum instantaneous velocity. These movement indexes allow for the determination of hypoxia thresholds of different fish species providing a novel approach for studying hypoxia indicators in fish. In conclusion, this research holds both theoretical and practical significance for the study of early hypoxia behavior in fish.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"232 ","pages":"Article 110079"},"PeriodicalIF":8.9000,"publicationDate":"2025-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"SDYOLO-Tracker: An efficient multi-fish hypoxic behavior recognition and tracking method\",\"authors\":\"Jiaxuan Yu , Guangxu Wang , Xin Li , Zhuangzhuang Du , Wenkai Xu , Muhammad Akhter , Daoliang Li\",\"doi\":\"10.1016/j.compag.2025.110079\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<div><div>Real-time monitoring and tracking of fish in aquaculture and timely detection of fish hypoxia behavior play a crucial role in advancing the intelligent aquaculture. This study proposes an effective multi-fish behavior identification and tracking method (SDYOLO-Tracker) to solve the lag problem of hypoxia behavior monitoring in different species of fish. Our work aims to enable early monitoring of abnormal hypoxic behavior in fish. To enhance the model ability to detect small objects, we improved the YOLOv8n model using SPD-Conv and D-LKA modules. Then, we integrate the improved YOLOv8n (SDYOLOv8) model with Bytetrack multi-object tracking (MOT) algorithm. This approach utilizes a complementary 3D motion prediction strategy, which effectively addresses issues such as sudden fish motion and disappearance, resulting in an accurate multi-fish identification and tracking. The experimental results indicate that the enhanced MOTA, HOTA, IDR and IDF1 have shown improvement of 3.32 %, 3.53 %, 6.68 %, and 5.36 %, respectively when compared with the original YOLOv8n model. Moreover, in the tested video, the ID number switching is reduced by 18.75 %, significantly improving the accuracy of multi-fish tracking without compromising the speed of the model. In the comparative experiments with other MOT algorithms, this method achieved the highest IDR and IDF1 metrics, with a frame per second (FPS) of 45.03, demonstrating the best tracking stability and the fastest model processing speed. In addition, SDYOLO-Tracker can conduct qualitative and quantitative analyses of fish behavior under varying dissolved oxygen concentrations, reflecting changes in locomotion, average velocity, and maximum instantaneous velocity. These movement indexes allow for the determination of hypoxia thresholds of different fish species providing a novel approach for studying hypoxia indicators in fish. In conclusion, this research holds both theoretical and practical significance for the study of early hypoxia behavior in fish.</div></div>\",\"PeriodicalId\":50627,\"journal\":{\"name\":\"Computers and Electronics in Agriculture\",\"volume\":\"232 \",\"pages\":\"Article 110079\"},\"PeriodicalIF\":8.9000,\"publicationDate\":\"2025-05-01\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Computers and Electronics in Agriculture\",\"FirstCategoryId\":\"97\",\"ListUrlMain\":\"https://www.sciencedirect.com/science/article/pii/S0168169925001851\",\"RegionNum\":1,\"RegionCategory\":\"农林科学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"2025/2/19 0:00:00\",\"PubModel\":\"Epub\",\"JCR\":\"Q1\",\"JCRName\":\"AGRICULTURE, MULTIDISCIPLINARY\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Computers and Electronics in Agriculture","FirstCategoryId":"97","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S0168169925001851","RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"2025/2/19 0:00:00","PubModel":"Epub","JCR":"Q1","JCRName":"AGRICULTURE, MULTIDISCIPLINARY","Score":null,"Total":0}
SDYOLO-Tracker: An efficient multi-fish hypoxic behavior recognition and tracking method
Real-time monitoring and tracking of fish in aquaculture and timely detection of fish hypoxia behavior play a crucial role in advancing the intelligent aquaculture. This study proposes an effective multi-fish behavior identification and tracking method (SDYOLO-Tracker) to solve the lag problem of hypoxia behavior monitoring in different species of fish. Our work aims to enable early monitoring of abnormal hypoxic behavior in fish. To enhance the model ability to detect small objects, we improved the YOLOv8n model using SPD-Conv and D-LKA modules. Then, we integrate the improved YOLOv8n (SDYOLOv8) model with Bytetrack multi-object tracking (MOT) algorithm. This approach utilizes a complementary 3D motion prediction strategy, which effectively addresses issues such as sudden fish motion and disappearance, resulting in an accurate multi-fish identification and tracking. The experimental results indicate that the enhanced MOTA, HOTA, IDR and IDF1 have shown improvement of 3.32 %, 3.53 %, 6.68 %, and 5.36 %, respectively when compared with the original YOLOv8n model. Moreover, in the tested video, the ID number switching is reduced by 18.75 %, significantly improving the accuracy of multi-fish tracking without compromising the speed of the model. In the comparative experiments with other MOT algorithms, this method achieved the highest IDR and IDF1 metrics, with a frame per second (FPS) of 45.03, demonstrating the best tracking stability and the fastest model processing speed. In addition, SDYOLO-Tracker can conduct qualitative and quantitative analyses of fish behavior under varying dissolved oxygen concentrations, reflecting changes in locomotion, average velocity, and maximum instantaneous velocity. These movement indexes allow for the determination of hypoxia thresholds of different fish species providing a novel approach for studying hypoxia indicators in fish. In conclusion, this research holds both theoretical and practical significance for the study of early hypoxia behavior in fish.
期刊介绍:
Computers and Electronics in Agriculture provides international coverage of advancements in computer hardware, software, electronic instrumentation, and control systems applied to agricultural challenges. Encompassing agronomy, horticulture, forestry, aquaculture, and animal farming, the journal publishes original papers, reviews, and applications notes. It explores the use of computers and electronics in plant or animal agricultural production, covering topics like agricultural soils, water, pests, controlled environments, and waste. The scope extends to on-farm post-harvest operations and relevant technologies, including artificial intelligence, sensors, machine vision, robotics, networking, and simulation modeling. Its companion journal, Smart Agricultural Technology, continues the focus on smart applications in production agriculture.