Deteksi Pejalan Kaki Pada Zebra Cross Untuk Peringatan Dini Pengendara Mobil Menggunakan Mask R-CNN

Wicaksono, Agung (2021) Deteksi Pejalan Kaki Pada Zebra Cross Untuk Peringatan Dini Pengendara Mobil Menggunakan Mask R-CNN. Undergraduate thesis, Institut Teknologi Sepuluh Nopember.

[img] Text
07211740000002-Undergraduate_Thesis.pdf - Accepted Version
Restricted to Repository staff only until 1 October 2023.

Download (9MB) | Request a copy

Abstract

Dewasa ini, fitur keselamatan pada kendaraan roda empat atau mobil sudah sangat berkembang pesat. Hal tersebut terbukti dengan banyaknya produsen mobil yang menerapkan teknologi seat belt, air bag, adaptive cruise control, electronic stability control, autonomous emergency braking, blind spot monitoring dan lain sebagainya. Namun, fitur yang sudah disebutkan diatas dinilai masih kurang ramah bagi pejalan kaki. Terbukti menurut data dari WHO, terdapat 270.000 pejalan kaki meninggal dunia setiap tahun atau sekitar 22% dari seluruh korban meninggal akibat kecelakan di jalan. Berawal dari permasalahan tersebut, penulis akan melakukan penelitian mengenai pendeteksian pejalan kaki pada zebra cross untuk peringatan dini pengendara mobil sebagai topik penelitian. Pada tugas akhir ini, terdapat 3 objek yang akan dideteksi yaitu pejalan kaki, zebra cross dan pengendara motor dengan menggunakan metode Mask R-CNN. Hasil terbaik yang didapatkan adalah pada penggunaan ResNet-101 untuk backbone Mask R-CNN dengan skor mAP sebesar 90.476%, mAR sebesar 88.889% serta F1-Score sebesar 87.777%, ================================================================================================ Today, safety features on four-wheeled vehicles or cars have developed very rapidly. This is evidenced by the number of car manufacturers that apply seat belt technology, air bags, adaptive cruise control, electronic stability control, autonomous emergency braking, blind spot monitoring and so on. However, the features mentioned above are still considered less friendly for pedestrians. It is proven that according to data from the WHO, there are 270,000 pedestrians who die every year or about 22% of all victims die due to road accidents. Starting from these problems, the author will conduct research on the detection of pedestrians at zebra cross for early warning car drivers as a research topic. In this final project, there are 3 objects to be detected, namely pedestrians, zebra cross and motorcyclists using the Mask R-CNN method. The best results obtained are the use of ResNet-101 for backbone Mask R-CNN with a score of mAP of 90.476%, mAR of 88.889% and F1-Score of 87.777%

Item Type: Thesis (Undergraduate)
Uncontrolled Keywords: Pedestrian, Zebra Cross, Mask R-CNN, Image Processing, Pejalan Kaki, Pengolahan Citra
Subjects: Q Science > Q Science (General) > Q325.5 Machine learning.
Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science)
Divisions: Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Computer Engineering > 90243-(S1) Undergraduate Thesis
Depositing User: Agung Wicaksono
Date Deposited: 02 Sep 2021 16:26
Last Modified: 02 Sep 2021 16:26
URI: https://repository.its.ac.id/id/eprint/91382

Actions (login required)

View Item View Item