Identifikasi Kemasan Produk FMCG Berdasarkan Karakteristik Warna Menggunakan Metode Hybrid Histogram HSV Support Vector Machine (SVM)

Sucahyo, Ilham (2026) Identifikasi Kemasan Produk FMCG Berdasarkan Karakteristik Warna Menggunakan Metode Hybrid Histogram HSV Support Vector Machine (SVM). Diploma thesis, Institut Teknologi Sepuluh Nopember.

Text 2040221087_Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (2MB) | Request a copy

Abstract

Dalam proses pemilahan produk secara manual dengan jumlah lima varian berbeda masih terjadi ketidaktelitian akibat human error. Akibatnya, tercatat sudah 10 kali kejadian tercampurnya varian yang tidak sejenis sehingga menghambat proses distribusi. Penelitian ini mengusulkan proses otomasi menggunakan image processing dengan memanfaatkan kamera webcam sebagai alat akuisisi citra yang terintegrasi dengan mekanisme sortir otomatis. Penelitian ini bertujuan mengembangkan sistem sortir otomatis untuk memisahkan lima varian produk sabun pada industri Fast Moving Consumer Goods (FMCG) yang sebelumnya mengalami permasalahan pencampuran produk akibat proses pemilahan secara manual. Proses klasifikasi dilakukan menggunakan metode ekstraksi fitur Histogram HSV dengan total 32 bin yang terdiri atas 16 bin kanal Hue, 8 bin kanal Saturation, dan 8 bin kanal Value, kemudian diklasifikasikan menggunakan algoritma Support Vector Machine (SVM). Hasil klasifikasi dikirimkan ke mikrokontroler ESP32 melalui komunikasi Wi-Fi untuk mengendalikan motor servo sebagai aktuator mekanisme sortir. Metode penelitian meliputi pengumpulan dataset, pra-pemrosesan citra, ekstraksi fitur, pelatihan model SVM, integrasi perangkat keras, serta pengujian sistem secara realtime. Hasil penelitian menunjukkan bahwa model terbaik diperoleh menggunakan pembagian dataset 70:20:10 dengan kombinasi hyperparameter kernel RBF, C = 10, dan gamma = scale, yang menghasilkan akurasi cross validation sebesar 99,71%, akurasi data testing sebesar 99,50%, serta akurasi data validation sebesar 97%. Pada implementasi sistem, model mampu mencapai akurasi klasifikasi realtime sebesar 90% dan berhasil diintegrasikan dengan mekanisme sortir otomatis berbasis ESP32 sehingga seluruh tahapan identifikasi dan penyortiran produk dapat berjalan secara realtime.
============================================================================================================================================
In the manual sorting process involving five different product variants, sorting inaccuracies caused by human error still occur. As a result, ten incidents of mixed product variants have been recorded, leading to delays in the distribution process. This study proposes an automated sorting system based on image processing using a webcam as the image acquisition device integrated with an automatic sorting mechanism. The objective of this research is to develop an automatic sorting system capable of separating five soap product variants in the Fast Moving Consumer Goods (FMCG) industry, where product mixing has previously occurred due to manual sorting. The classification process employs the HSV color histogram feature extraction method with a total of 32 histogram bins consisting of 16 Hue bins, 8 Saturation bins, and 8 Value bins, followed by classification using the Support Vector Machine (SVM) algorithm. The classification results are transmitted to an ESP32 microcontroller via Wi-Fi communication to control servo motors as actuators of the sorting mechanism. The research methodology includes dataset collection, image preprocessing, feature extraction, SVM model training, hardware integration, and real-time system evaluation. The experimental results show that the best-performing model was obtained using a 70:20:10 train-test-validation split with the RBF kernel, C = 10, and gamma = scale, achieving a cross-validation accuracy of 99.71%, a testing accuracy of 99.50%, and a validation accuracy of 97%. Furthermore, the proposed system achieved a real-time classification accuracy of 90% and was successfully integrated with the automatic sorting mechanism, enabling the entire identification and sorting process to operate automatically in real time.

Item Type:	Thesis (Diploma)
Uncontrolled Keywords:	Kata kunci: Support Vector Machine (SVM), Histogram HSV, Internet of Things, Machine Learning, Mekanisme Sortir. ======================================================================================================================== Keyword: Support Vector Machine (SVM), Histogram HSV, Internet of Things, Machine Learning, Mekanisme Sortir.
Subjects:	T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK5105.546 Computer algorithms T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK6592.A9 Automatic tracking. T Technology > TS Manufactures > TS195 Packaging
Divisions:	Faculty of Vocational > 36304-Automation Electronic Engineering
Depositing User:	Ilham Edo Sucahyo
Date Deposited:	05 Aug 2026 04:45
Last Modified:	05 Aug 2026 04:45
URI:	http://repository.its.ac.id/id/eprint/143964

Actions (login required)

View Item