RANCANG BANGUN SISTEM PENILAIAN KUALITAS CUCI TANGAN BERBASIS VIDEO SECARA WAKTU-NYATA DENGAN CONVOLUTIONAL NEURAL NETWORK (CNN)

DAFFA AJI PANGESTU, . (2023) RANCANG BANGUN SISTEM PENILAIAN KUALITAS CUCI TANGAN BERBASIS VIDEO SECARA WAKTU-NYATA DENGAN CONVOLUTIONAL NEURAL NETWORK (CNN). Sarjana thesis, UNIVERSITAS NEGERI JAKARTA.

	Text COVER.pdf Download (593kB)
	Text BAB 1.pdf Download (244kB)
	Text BAB 2.pdf Restricted to Registered users only Download (673kB) | Request a copy
	Text BAB 3.pdf Restricted to Registered users only Download (2MB) | Request a copy
	Text BAB 4.pdf Restricted to Registered users only Download (1MB) | Request a copy
	Text BAB 5.pdf Restricted to Registered users only Download (390kB) | Request a copy
	Text DAFTAR PUSTAKA.pdf Download (1MB)
	Text LAMPIRAN.pdf Restricted to Registered users only Download (1MB) | Request a copy

Abstract

Proses mencuci tangan sangat efektif dalam pencegahan penyebaran penyakit. Namun, evaluasi kualitas cuci tangan yang akurat dan objektif masih menjadi tantangan hingga kini. Oleh karena itu diusulkan perancangan perangkat dan implementasi sebuah sistem penilaian kualitas cuci tangan berbasis video secara waktu-nyata untuk memberikan solusi otomatis dalam mengevaluasi kualitas cuci tangan. Metode penelitian melibatkan pengumpulan dataset video yang mencakup variasi gerakan cuci tangan menggunakan perangkat kamera yang terintegrasi pada sistem Raspberry-Pi. Dataset berupa citra dengan filter warna skin mask digunakan untuk melatih dan menguji model Convolutional Neural Network (CNN). Model CNN mengekstraksi fitur penting dan mengklasifikasikan citra sesuai dengan 6 katagori gerakan pencucian tangan menurut World Health Organization (WHO). Hasil eksperimen model CNN menunjukkan nilai akurasi rata-rata diatas 96% yang dilatih dengan arsitektur MobileNet, MobileNetV2, DenseNet121, NASNetMobile, ResNet50, dan VGG19. Selain itu, sistem ini juga mampu berjalan secara waktu-nyata dengan nilai frame per second maksimum sebesar 45 fps. Sistem juga akan memberikan umpan balik berupa informasi persentase kualitas pencucian tangan yang dinilai berdasarkan durasi dan kelengkapan gerakan pencucian tangan. Penuruan nilai akurasi terjadi saat pengujian sistem akibat variasi data baru dan kondisi lingkungan yang tidak terkendali. **** The handwashing process is very effective in preventing the spread of disease. However, accurate and objective evaluation of handwashing quality remains a challenge. Therefore it is proposed to design a device and implement a video-based handwashing quality assessment system in real-time to provide an automatic solution for evaluating the quality of handwashing. The research method involves collecting a video dataset that includes variations of handwashing movements using a camera integrated into the Raspberry-Pi system. The dataset in the form of images with skin mask color segmentation is used to train and test the (CNN) Convolutional Neural Network model. The CNN model will extract essential fiturs and classify images according to six categories of handwashing movements, according to the World Health Organization (WHO). The experimental results of the CNN model show an average accuracy value of above 96% which is trained with the MobileNet, MobileNetV2, DenseNet121, NASNetMobile, ResNet50, and VGG19 architectures. In addition, this system is also capable of running in real-time with maximum frame-per-second value of 45 fps. The system will also provide feedback in the form of information on the proportion of handwashing quality assessed based on the duration and completeness of the handwashing movements. A decrease in the value of accuracy occurs when testing the system due to variations in new data and uncontrolled environmental conditions.

Item Type:	Thesis (Sarjana)
Additional Information:	1). dr. rer. nat. Bambang Heru Iswanto, M.Si 2). Dr. Esa Prakasa, M.T
Subjects:	Sains > Matematika > Software, Sistem Informasi Komputer Sains > Fisika
Divisions:	FMIPA > S1 Fisika
Depositing User:	Users 18972 not found.
Date Deposited:	01 Sep 2023 06:53
Last Modified:	01 Sep 2023 06:53
URI:	http://repository.unj.ac.id/id/eprint/39975

Actions (login required)

View Item