SHELL TYPE CLASSIFICATION SYSTEM BASED ON SHELL IMAGES USING SUPPORT VECTOR MACHINE (SVM)

Abstract

This study aims to build an automatic classification system to identify shellfish types based on shell images by applying the Support Vector Machine (SVM) algorithm. This study classifies three types of shellfish, namely blood cockles with the scientific name Anadara granosa, green mussels (Perna viridis), and scallops (Amusium pleuronectes). Image data was obtained from the internet and each class consisted of 150 images, so the total dataset was 450 images. The research stages include image pre-processing to normalize image size and quality, feature extraction to obtain visual information in the form of texture (with GLCM), color (RGB histogram), and shape (Canny edge detection), and classification using SVM. This application is web-based and functions to receive uploaded shellfish images from users and provide automatic shellfish type recognition results. The test results show that the developed SVM model is able to classify shellfish types with high accuracy, reaching 93,83%. This research is expected to contribute to the development of digital shellfish species identification technology to support the fields of fisheries, marine resource conservation, and marine biota research.