Pengembangan YOLOv3 Dengan Fitur Ekstraktor MobileNetv2 Untuk Deteksi Dan Klasifikasi Kendaraan Bergerak

Sholehurrohman, Ridho (2021) Pengembangan YOLOv3 Dengan Fitur Ekstraktor MobileNetv2 Untuk Deteksi Dan Klasifikasi Kendaraan Bergerak. Masters thesis, Institut Teknologi Sepuluh Nopember.

	Text 06111950010001-Master_Thesis.pdf Restricted to Repository staff only until October 2023. Download (7MB) | Request a copy
	Text 06111950010001-Master_Thesis.pdf Restricted to Repository staff only Download (7MB) | Request a copy
	Text 06111950010001-Master_Thesis.pdf Restricted to Repository staff only Download (7MB) | Request a copy
	Text 06111950010001-Master_Thesis.pdf Restricted to Repository staff only Download (7MB) | Request a copy
	Text 06111950010001-Master_Thesis.pdf Restricted to Repository staff only Download (7MB) | Request a copy

Abstract

YOLOv3-Darknet53 memiliki akurasi deteksi yang baik, namun kurang dalam kecepatan waktu. Untuk meningkatkan performansi dari sisi komputasi digunakan fitur ekstraktor MobileNetv2. Pada penelitian ini telah dikembangkan arsitektur YOLOv3 dengan fitur ektsraktor MobileNetv2 untuk deteksi dan klasifikasi kendaraan bergerak. Pengembangan arsitektur dilakukan dengan beberapa tahap, di antaranya adalah penyesuaian convolusi, menghilangkan layer Fully Connected dan Softmax serta menambahkan beberapa layer. Dari arsitektur yang dikembangkan telah berhasil dirancang sebuah algoritma YOLOv3-MobileNetv2 yang mempunyai kompleksitas O(km^2) yang lebih baik dibandingkan dengan kompleksitas algoritma YOLOv3-Darknet53 yaitu O(km^2n^2). Berdasarkan hasil uji coba sistem didapat waktu pemrosesan dari YOLOv3-MobileNetv2 untuk melakukan deteksi dan klasifikasi kendaraan bergerak adalah 0,124150 second untuk setiap frame sedangkan waktu pemrosesan dari YOLOv3-Darknet53 adalah 0,42633 second untuk setiap frame. Hal tersebut menunjukkan bahwa YOLOv3-MobileNetv20,130218 second per frame lebih cepat dibandingkkan YOLOv3-Darknet53. Didapatkan juga nilai akurasi rata-rata YOLOv3-MobileNetv2 94,84% dan YOLOv3-Darknet53 90,52%.

============================================================================YOLOv3-Darknet53 has good detection accuracy but badly in time speed. In order to improved from the computing side used MobileNetv2 feature extraction. The YOLOv3 architecture has been developed with the MobileNetv2 feature extractotion for the detection and classification of moving vehicle. Architectural development was carried out in several stages, including convolution adjustment, eliminated Fully Connected and Softmax layers, and adding several layers. From the architecture developed has been successfully designed the YOLOv3-MobileNetv2 algorithm that has a complement of O(km^2) which is better compared to the complexity than the YOLOv3-Darknet53 algorithm that is O(km^2n^2). Based on the results of the system trials obtained processing time from YOLOv3-MobileNetv2 to detection and classification moving vehicles is 0,12415 second for each single frame while the processing time of YOLOv3-Darknet53 is 0,42633 second for each single frame. This shows that YOLOv3-MobileNetv2 0.30218 second per frame is faster than YOLOv3-Darknet53. The average accuracy score of YOLOv3-MobileNetv2 is 94,84% and YOLOv3-Darknet53 is 90,52%.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	Kecerdasan Buatan, YOLOv3, Darknet53, MobileNetv2, Deteksi dan Klasifikasi Objek ====== Artificial Intelligence, YOLOv3, Darknet53, MobileNetv2, Detection and Classification Object
Subjects:	Q Science > Q Science (General) > Q325.5 Machine learning. Q Science > QA Mathematics > QA336 Artificial Intelligence Q Science > QA Mathematics > QA75 Electronic computers. Computer science. EDP Q Science > QA Mathematics > QA76.6 Computer programming. Q Science > QA Mathematics > QA76.625 Internet programming. Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science) Q Science > QA Mathematics > QA76.9 Computer algorithms. Virtual Reality. Computer simulation. Q Science > QA Mathematics > QA76.9.I52 Information visualization T Technology > TA Engineering (General). Civil engineering (General) > TA1637 Image processing--Digital techniques
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Mathematics > 44101-(S2) Master Thesis
Depositing User:	Ridho Sholehurrohman
Date Deposited:	01 Sep 2021 03:14
Last Modified:	01 Sep 2021 03:14
URI:	http://repository.its.ac.id/id/eprint/90714

Actions (login required)

View Item