Fish Detection Based on Halcon and MobileNetV2 for Automatic Fish Feeding System

Abstract

Fish detection is an important technology to achieve intelligent, accurate and appropriate feeding of fish. The detection rate of traditional detection methods is low and the state of the fish school will have a certain impact on the accuracy of the feeding strategy. In order to improve the recognition rate and obtain the state of fish school, a low-resolution image fish detection method based on Halcon software and MobileNetV2 network is designed. The position coordinates and head orientation of fish are extracted to analyze the movement characteristics of fish school by MobileNetV2 network. The experimental results show that the detection method has a high recognition rate for fish detection. This method provides an important reference value for the design of automatic fish feeding system.