Rancang Bangun Sistem Pengukuran Warna Berbasis Computer Vision Untuk Quality Control Produk Minyak Goreng

Khoiroh, Siti Ni`matul (2025) Rancang Bangun Sistem Pengukuran Warna Berbasis Computer Vision Untuk Quality Control Produk Minyak Goreng. Diploma thesis, Institut Teknologi Sepuluh Nopember.

Text 2042221048-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (6MB) | Request a copy

Abstract

Budidaya kelapa sawit telah dilakukan secara meluas di Indonesia. Sehingga, total produksi minyak kelapa sawit indonesaia mencapai 56,5 juta ton. CPO dapat diolah menjadi beberapa produk salah satunya adalah minyak goreng. Dalam proses produksi minyak goreng, warna berhubungan dengan proses bleaching. Untuk menjaga konsistensi produk perlu dilakukan inspeksi warna. Saat ini metode inspeksi warna masih dilakukan secara konvensional dimana dalam penerapannya membutuhkan biaya yang mahal, waktu lama dan hasil tidak konsisten. Oleh karena itu, dilakukanlah perancangan sistem pengukuran warna menggunakan computer vision. Computer vision mempunyai kelebihan diantaranya biaya yang dibutuhkan relatif terjangkau, proses komputasi cepat dan pengukuran tidak perlu kontak langsung dengan objek. Komponen yang digunakan dalam peracangan computer vision pada proyek akhir ini meliputi Raspberry PI 5, Webcam Logitech c922, LED dan LCD Waveshare 4.3”. Pengambilan gambar dilakukan oleh webcam. Hasil gambar dikonversi dari RGB ke CIELab untuk menselaraskan pengelihatan computer vision dengan manusia. Nilai a* (Red-Green) dan b*(Blue-Yellow) diklasifikasikan menggunakan K-Nearest Neighbour (KNN) dan Naïve Bayes classifier untuk menentukan jenis warna sesuai dengan standar SNI 7709:2019. Dalam proses testing menggunakan dataset, model K-NN mempunyai nilai akurasi 1, presisi 1, recall 1 dan F1-Score 1. Sedangkan model Naive Bayes mempunyai nilai akurasi 0.59, presisi 0.62, recall 0.60 dan F1-Score 0.56. Ketika dilakukan pengujian fungsionalitas model K-NN mempunyai nilai evaluasi matriks yang tinggi. Namun, hasil klasifikasi tidak sesuai dengan kelas sampel sebenarnya. Sedangkan model naive bayes, nilai evaluasi metrik yang dihasilkan rendah tetapi model melakukan klasifikasi sesuai dengan label sebenarnya dan hasilnya konsisten.
==============================================================================================================================
Oil palm cultivation has been widely cultivated in Indonesia. Consequently, palm oil production has reached 56.5 million tons. CPO can be processed into several products, one of them is cooking oil. In the cooking oil production process, color is related with bleaching process. To maintain product consistency, color inspection is necessary. Currently, color inspection is performed conventionally, which is expensive, time-consuming, and produces inconsistent results. Therefore, a color measurement system was designed using computer vision. Computer vision offers advantages such as relatively low costs, fast computational processes, and non-contact measurement. The components used in this final project include Raspberry PI 5, Logitech c922 Webcam, LED and 4.3” Waveshare LCD. Image is captureed using Webcam. The resulting image is converted from RGB to CIELab to align computer vision with human vision. The a* (Red-Green), b* (Blue-Yellow) values are classified using K-Nearest Neighbor (KNN) and Naïve Bayes classifier to determine the type of color based pn SNI 7709:2019. During the testing process using the dataset, the K-NN model achieved an accuracy, precision, recall, and F1-score value of 1.00. In contrast, the Naive Bayes model yielded values of 0.59 for accuracy, 0.62 for precision, 0.60 for recall, and 0.56 for the F1-score. Functional testing of the K-NN model revealed high evaluation metrics. However, the classification results did not align with the actual sample classes. Meanwhile, Naive Bayes model produced lower evaluation metrics, but classified samples according to their actual labels, and the results were consistent.

Item Type:	Thesis (Diploma)
Uncontrolled Keywords:	Minyak Goreng, Computer Vision, CIELab, K-NN, Naïve Bayes, Cooking Oil, Computer Vision, CIELab, K-NN, Naïve Bayes.
Subjects:	T Technology > TA Engineering (General). Civil engineering (General) > TA1637 Image processing--Digital techniques. Image analysis--Data processing.
Divisions:	Faculty of Vocational > Instrumentation Engineering
Depositing User:	Siti Ni`matul Khoiroh
Date Deposited:	31 Jul 2026 07:46
Last Modified:	31 Jul 2026 07:46
URI:	http://repository.its.ac.id/id/eprint/140839

Actions (login required)

View Item