A Copula Based Method for the Classification of Fish Species

Abstract

The proposed work develops a method for classification of the species of a fish given in an image, which is a sub-ordinate level classification problem. Fish image categorization is unique and challenging as the images of same fish species can show significant differences in the fish's attributes when taken in different conditions. The authors' approach analyses the local patches of images, cropped based on specific body parts, and hence keep comparison more specific to grab more finer details rather than comparing global postures. The authors have used Histogram of Oriented Gradients and colour histograms to create representative feature vectors; feature vectors are summarized using Copula theory. Their method is very simple yet they have matched the classification accuracy of other proposed complex work for such problems.

SD-YOLOv8: An Accurate Seriola dumerili Detection Model Based on Improved YOLOv8

“…Cutter et al [ 19 ] implemented a cascade of Haar features for the detection and recognition of benthic fish during unconstrained underwater surveys. Dhawal and Chen [ 20 ] generated representative feature vectors through the use of a histogram of oriented gradients (HOG) and color histograms for the identification of 10 similar fish species. While these appearance-based techniques have shown commendable detection performance in static images, they require considerable human effort in feature design and are less robust in adapting to dynamic and complex underwater environments, exacerbated by limited data availability.…”