Abstract:

In intensive aquaculture, the real-time detection and monitoring of common infectious disease is an important basis for scientific fish epidemic prevention strategies that can effectively reduce fish mortality and economic loss. However, low-quality underwater images and low-identification targets present great challenges to diseased fish detection. To overcome these challenges, this paper proposes a diseased fish detection model, using an YOLOV5 network for aquaculture . wide range of applications in marine ecology extensively uses underwater cameras. Still, to efficiently process the vast amount of data generated, we need to develop tools that can automatically detect and recognize species captured on film. Classifying fish species from videos and images in natural environments can be challenging because of noise and variation in illumination and the surrounding habitat. In this paper, we propose a two-step deep learning approach for the detection and classification of temperate fishes with-out pre-_filtering. The first step is to detect each single fish in an image, independent of species and sex. For this purpose, we employ the You Only Look Once(YOLO) object detection technique. In the second step,we adopt a Convolutional Neural Network (CNN). (Fish4Knowledge), whereupon both the object detection and classifier are updated with temperate fishes of interest.
Keyword: Biometric Fish Classification ,collected dataset,YOLO V5 ,CNN

INTRODUCTION:

Coastal marine ecosystems provide habitats for spawning, nursing, and feeding for a diverse fish community.Due to the highly complex and dynamic nature of this environment, it is challenging to monitor and study ecological processes . High resolution underwater camera technologies have recently made it possible to obtain large volumes of observations from remote areas and allowed for better capture the species' cryptic behavior and changes in the environment. Although comprehensive image and video data can be collected, the processing is of image data in ecological context is mostly manual and therefore very labor-intensive . As a result, only a portion of the available recordings can be analyzed which is greatly limiting the potential advances that can be made from these data streams. Furthermore, the accuracy of human-based visual assessments are highly dependent on conditions of the underwater environment and taxonomic expertise in interpreting the data . Therefore, an objective analytical tool capable of processing image data fast and efficient is most welcomed by scientists and resource management. To release the burden of manual processing, and to improve the classi_cation accuracy, computer vision- based approaches have increasingly been employed in marine ecology analysis
China is the largest consumer of freshwater fish in the world . Wild-caught freshwater fish make up a large portion of freshwater fish consumption . Overfishing due to consumption of wild-caught freshwater fish destroys biodiversity . In order to reduce the ecological damage caused by fishing, in 2020, the Chinese Ministry of Agriculture and Rural Affairs announced the start of a 10-year fishing ban on the Yangtze River . With the implementation of the fishing ban, China’s freshwater fish consumer market is more dependent on artificial culture. During freshwater aquaculture, fish feed or fertilizer is put into the water to increase freshwater fish production, but residual fertilizer, fish feces and other excreta can cause water eutrophication and lead to eco-catastrophe, such as red tides, so this type of freshwater aquaculture was banned in natural waters. In this context, the proportion of intensive aquaculture will further increase . One study shows that intensive aquaculture systems will have to predominate .