Analisis Perbandingan Model CNN Terhadap Klasifikasi Citra Komponen Elektronika

Arrosyid, Muhammad Zydane
Hermawan, Arief
., Sutarman

Algoritme Jurnal Mahasiswa Teknik Informatika
Vol 5 No 2 (2025): April 2025 || Algoritme Jurnal Mahasiswa Teknik Informatika

This study compares various Convolutional Neural Network (CNN) models in classifying electronic component images. The background of this research stems from the need to automatically identify and classify components in environments with limited computational resources. The data used in this research was collected through image scraping from the internet, supplemented by direct image acquisition using a camera. The data was then processed and trained using several CNN models, including MobileNet, NASNetLarge, VGG16, and others, as well as a custom CNN model developed by the researcher. The results show that NASNetLarge achieved the highest test accuracy of 79.31%, while MobileNet demonstrated high efficiency in computational resource usage. This study highlights that model size does not always correlate with accuracy, and models with fewer parameters can provide effective solutions for resource-constrained conditions.