Classifying the Swallow Nest Quality Using Support Vector Machine Based on Computer Vision

Abstract

Swallow Nest is a valuable export commodity, particularly in Indonesia. It is produced when a swallow's saliva hardens and is frequently encountered in high-rise buildings. Swallow nests can be utilized to treat various ailments in the medical sector. The price of a swallow nest varies according to its quality, which is commonly classified into three grades: quality 1 (Q1), quality 2 (Q2), and quality 3 (Q3). Q1 is of the highest quality, while Q3 is of the lowest. Each grade has a different physical appearance. Currently, many people lack knowledge regarding the grade of a swallow nest. Therefore, a method is needed to automatically classify the quality of swallow nests based on computer vision. The proposed method consists of several main processes, including image acquisition, ROI detection, pre-processing, segmentation, feature extraction, and classification. The feature extraction was applied based on shapes, followed by the Support Vector Machine (SVM) implementation in the classification process. This process was performed with cross-validation using the k-fold values of 5. The performance evaluation was done using three parameters: precision, recall, and accuracy, by achieving the value of 90.6%, 89.3%, and 89.3%, respectively.