2018/07/16 by Ranju Mandal, Rod M. Connolly, Mandal, Ranju +5
Biochemistry, Genetics and Molecular Biology · Environmental Science · #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #Ichthyology and Marine Biology #Identification and Quantification in Food #Machine Learning (cs.LG) #Water Quality Monitoring Technologies
paper · pdf · doi:10.48550/arxiv.1807.05838
openalex publication_date 2018/07/16 · openalex created_date 2022/08/04 · openalex updated_date 2026/07/28
Uses of underwater videos to assess diversity and abundance of fish are being\nrapidly adopted by marine biologists. Manual processing of videos for\nquantification by human analysts is time and labour intensive. Automatic\nprocessing of videos can be employed to achieve the objectives in a cost and\ntime-efficient way. The aim is to build an accurate and reliable fish detection\nand recognition system, which is important for an autonomous robotic platform.\nHowever, there are many challenges involved in this task (e.g. complex\nbackground, deformation, low resolution and light propagation). Recent\nadvancement in the deep neural network has led to the development of object\ndetection and recognition in real time scenarios. An end-to-end deep\nlearning-based architecture is introduced which outperformed the state of the\nart methods and first of its kind on fish assessment task. A Region Proposal\nNetwork (RPN) introduced by an object detector termed as Faster R-CNN was\ncombined with three classification networks for detection and recognition of\nfish species obtained from Remote Underwater Video Stations (RUVS). An accuracy\nof 82.4% (mAP) obtained from the experiments are much higher than previously\nproposed methods.\n