Fish Detection and Classification

Abstract

Fish detection and classification is a very difficult and complex task due to its commercialization and agriculture importance. There are lots of problems that arise while detection like segmentation error, noise, distortion in image etc. To tackle these issues, various techniques have been used like K-Nearest Neighbour (KNN) and Support Vector Machine (SVM). The SVM method was used to identify fish species using four kernel functions: linear, polynomial, sigmoid, and radial substructure functions. A Support Vector Machine-based strategy is proposed for improving fish species detection and overcoming the limitations of various existing approaches. The act of knowing or differentiating fish species based on their traits is known as fish identification. It's also a method of matching fish species based on similar attributes in pictures of representative specimens. Patterns, determination of physical or behavioural attributes and contour matching, quality control, feature extraction and statistical analysis of fish species are all reasons why fish detection is required.