{"title":"水生动物物种自动识别与检测新方法","authors":"Pratik K.Agrawal, Vaishnavi Kamdi, Ishan Mittal, Pranav Bobde, Amarsingh Kashyap","doi":"10.47164/ijngc.v14i1.1013","DOIUrl":null,"url":null,"abstract":"Marine fisheries contribute greatly to the economic aspects of any country. India, having a coastline of almost 8000 KM, a surplus of fisheries potential could be estimated here. Because of this vast coastal area, active reporting of captured fishes is difficult through manual monitoring. Computer-aided approach is the best suitable option during the active season. This paper focuses on investigating an approach for identifying single as well as multiple aquatic animal species in a single image. Further a responsive web as well as mobile application are developed, in which the ML models are integrated. This will help users to access data as per their use. The method used YOLOv5n, a lightweight object detection algorithm, to detect these species. The trained model yielded mAP@0.5:0.95 intersection over union (IoU), and average precision (AP) for each species. The species’ AP varied as well. Few GFLOPs are used by YOLOv5n. This indicates that it is a scaled-down version capable of running on the 5.1 GFLOP Raspberry Pi 3B+. Despite employing substantially fewer GFLOPs, YOLOv5n outperformed Faster R-CNN.","PeriodicalId":42021,"journal":{"name":"International Journal of Next-Generation Computing","volume":"25 1","pages":""},"PeriodicalIF":0.3000,"publicationDate":"2023-02-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Novel Approach to Automatic Identification and Detection of Aquatic Animal Species\",\"authors\":\"Pratik K.Agrawal, Vaishnavi Kamdi, Ishan Mittal, Pranav Bobde, Amarsingh Kashyap\",\"doi\":\"10.47164/ijngc.v14i1.1013\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Marine fisheries contribute greatly to the economic aspects of any country. India, having a coastline of almost 8000 KM, a surplus of fisheries potential could be estimated here. Because of this vast coastal area, active reporting of captured fishes is difficult through manual monitoring. Computer-aided approach is the best suitable option during the active season. This paper focuses on investigating an approach for identifying single as well as multiple aquatic animal species in a single image. Further a responsive web as well as mobile application are developed, in which the ML models are integrated. This will help users to access data as per their use. The method used YOLOv5n, a lightweight object detection algorithm, to detect these species. The trained model yielded mAP@0.5:0.95 intersection over union (IoU), and average precision (AP) for each species. The species’ AP varied as well. Few GFLOPs are used by YOLOv5n. This indicates that it is a scaled-down version capable of running on the 5.1 GFLOP Raspberry Pi 3B+. Despite employing substantially fewer GFLOPs, YOLOv5n outperformed Faster R-CNN.\",\"PeriodicalId\":42021,\"journal\":{\"name\":\"International Journal of Next-Generation Computing\",\"volume\":\"25 1\",\"pages\":\"\"},\"PeriodicalIF\":0.3000,\"publicationDate\":\"2023-02-15\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"International Journal of Next-Generation Computing\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.47164/ijngc.v14i1.1013\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"International Journal of Next-Generation Computing","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.47164/ijngc.v14i1.1013","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
Novel Approach to Automatic Identification and Detection of Aquatic Animal Species
Marine fisheries contribute greatly to the economic aspects of any country. India, having a coastline of almost 8000 KM, a surplus of fisheries potential could be estimated here. Because of this vast coastal area, active reporting of captured fishes is difficult through manual monitoring. Computer-aided approach is the best suitable option during the active season. This paper focuses on investigating an approach for identifying single as well as multiple aquatic animal species in a single image. Further a responsive web as well as mobile application are developed, in which the ML models are integrated. This will help users to access data as per their use. The method used YOLOv5n, a lightweight object detection algorithm, to detect these species. The trained model yielded mAP@0.5:0.95 intersection over union (IoU), and average precision (AP) for each species. The species’ AP varied as well. Few GFLOPs are used by YOLOv5n. This indicates that it is a scaled-down version capable of running on the 5.1 GFLOP Raspberry Pi 3B+. Despite employing substantially fewer GFLOPs, YOLOv5n outperformed Faster R-CNN.