Improving Detection Performance of Helmetless Motorcyclists using the Combination of HOG, HOP, and LDB Descriptors
Penulis/Author
SUTIKNO (1); Prof. Drs. Agus Harjoko, M.Sc., Ph.D. (2); Afiahayati, S.Kom., M.Cs., Ph.D (3)
Tanggal/Date
2022
Kata Kunci/Keyword
Abstrak/Abstract
A traffic accident is one of the causes of death in the world and motorcyclists who do not wear helmets are certainly one type of likely victims of this. The number of death can be reduced by making a system capable of detecting helmetless motorcyclists. This study aimed to develop a system to detect helmetless motorcyclists. This system was divided into three subsystems of moving objects segmentation, motorcycle classification, and helmetless heads detection. The new descriptor proposed here was Histogram of Oriented Phase and Gradient - Local Difference Binary (HOPG-LDB) descriptor to improve the accuracy of motorcycle classification and helmetless heads detection. The HOPG-LDB descriptor was a combination of Histograms of Oriented Gradients (HOG), Histogram of Oriented Phase (HOP), and Local Difference Binary (LDB). The experiments were performed using a Multilayer Perceptron (MLP) classifier and 2 datasets of images that were taken from the front and rear of motorcycles. The experimental results show that the proposed new descriptor was capable of improving detection accuracy from a single descriptor and combinations of two descriptors for motorcycle classification, heads detection, and helmetless heads detection. The experimental result also shows that the descriptor we proposed produced higher accuracy than previous work.
Rumpun Ilmu
Ilmu Komputer
Bahasa Asli/Original Language
English
Level
Internasional
Status
Dokumen Karya
No
Judul
Tipe Dokumen
Aksi
1
2022022839.pdf
Bukti Published
2
IJIES Detection of Helmetless Motorcyclist-lengkap-PAK-low.pdf
