STRATIFICATION OF FISH SPECIES USING COMPARATIVE MACHINE LEARNING ALGORITHMS

The fisheries industry and aqua-ecology necessitates the precise stratification of fish species, and endeavours to agnize each type offish based on certain characteristics. Conventional methods engage in manual characterisation which may not be optimal interms of time and impeccable results. While there have been various strategies to analyse and identify the diverse fish species, the flaws of traditional approaches are surpassed through different domains such as machine learning, deep learning and Internet of Things (IoT). The proposed research pivots onstratifying fish species using comparative machine learning algorithmsto overcome the problems related to classifying similar species, while significantly ensuring accurate classification resultsin optimal time. This indagationutilizes the image viewer tool to unsheathe the characteristics of the procured fish type, while building the database to effectuate classification of the different species. The research incorporates the classification juxtapose through three machine learning algorithms such as the Cubic Support Vector Machines (CSVM), Cosine K-Nearest Neighbour (CosKNN) and the ensemble method using bagged trees and Ada boost optimizer toaccurately classify fish species based on morphological features extracted from images. The proposedmethodology involves the collection of a comprehensive dataset comprising images of different fish species, while sufficientlyextractingrelevant features, andfurther entailthe application of machine learning algorithms for classification.This research thus contributes to the advancement of automated fish species identification, which can profoundly augment the proficiency and verity of fisheries management, along with enhancing the conservation of aquatic ecosystems. The simulations are carried out in MATLAB, and the results evince the ensemble method yields thezenith of classification accuracy as compared to the other two algorithms.