Research On Aquaculture Security Methods Based On Behavior Recognition And Signal Analysis

Aquaculture is an important part of China’s national economy,which has greatly improved the income level of the people.With the innovation and development of China’s modern information technology,aquaculture technology has also been improved to a high level.However,due to the impact of bad theft,the safety of the breeding plant has always troubled the farmers,and has also brought a certain degree of economic loss to the breeding industry,which has increased the burden of breeding.In practical applications,intelligent monitoring,abnormal behavior detection,and suspicious target detection all depend on the development of image analysis technology.However,relying only on video image analysis has certain limitations,because it is difficult to avoid the effects of lighting,occlusion,angle,and position.In actual production,abnormal events are often accompanied by abnormal sounds,and sound signals also have the effect of information transfer.Therefore,the use of image analysis technology to identify suspicious behavior,coupled with the analysis of abnormal sound signals,has high practical value in theft detection in the breeding industry.This article focuses on the safety prevention and control technology in farms,focusing on methods for identifying abnormal behaviors and analyzing abnormal sounds.Since a good aquatic environment is the basis for the healthy growth of aquatic organisms,this article combines the theft detection method with the monitoring of the aquatic water environment to make the proposed method closer to the application scenario of the aquaculture industry.The main work of this article is as follows:First of all,for the establishment of a water environment monitoring system for farms using the Internet of Things technology,the water quality monitoring indicators in the system are determined according to the requirements of the culture.Set up a water quality information database for farmers,so that users can view the real situation of the farm in real time and take corresponding measures in a timely manner.Then,aiming at the problem that the number of theft abnormal behavior database samples is insufficient,we establish our own behavior database and use YOLO network to recognize the abnormal behavior.Compared with traditional behavior recognition methods based on feature extraction,neural network-based methods have better real-time performance.Finally,research on the analysis method of abnormal sound signals,aiming at the problem of insufficient existing sound samples,based on the existing sound data set,establish an abnormal sound database accompanying the theft,and perform feature extraction and analysis.The recognition results of abnormal sound signals with different features under different signal-to-noise ratios are obtained.The experimental results show that the detection of abnormal sound signals can be used as an auxiliary means to detect abnormal theft behavior and improve the reliability of safety prevention and control methods.