Pengaturan Waktu Penyeberangan Adaptif Berdasarkan Deteksi Objek Pedestrian Berbasis Convolutional Neural Network (Cnn)

Resha, Randy Putra (2019) Pengaturan Waktu Penyeberangan Adaptif Berdasarkan Deteksi Objek Pedestrian Berbasis Convolutional Neural Network (Cnn). Other thesis, Institut Teknologi Sepuluh Nopember.

Text 07211540000034-Undergraduate_Theses.pdf Restricted to Repository staff only Download (28MB) | Request a copy

Abstract

Berjalan kaki merupakan cara yang cepat untuk menyelesaikan perjalanan pendek di daerah urban. Berbagai fasilitas pendukung telah disediakan bagi pejalan kaki salah satunya infrastruktur fasilitas penyeberangan. Di Indonesia, berdasarkan data dari Korps Lalu Lintas Kepolisian pada 6 bulan terakhir kecelakaan pada pejalan kaki yaitu sejumlah 8309 orang, dan diantaranya 379 orang terlibat dalam kecelakaan pada fasilitas penyeberangan pada zebra-cross [3]. Dalam penggunaan fasilitas tersebut dinilai masih kurang maksimal karena kurang tepatnya alokasi waktu yang disediakan bagi pedestrian untuk menyeberang yang hingga saat ini berupa waktu yang tetap, sehingga kelalaian pengguna pedestrian yang menyeberang secara terburu-buru lebih tinggi. Kendala tersebut dapat diatasi dengan pemanfaatan teknologi deep learning object detection sebagai penentuan waktu penyeberangan bagi pedestrian berdasarkan analisa jumlah pedestrian dan pendeteksian adanya pedestrian yang sedang menggunakan pelican-crossing dengan penangkapan gambar atau video dari IP Camera. Dengan adanya sistem pengaturan waktu penyeberangan yang adaptif tersebut diharapkan dapat mengurangi terjadinya kelalaian dan peningkatan keamanan serta kenyamanan bagi pedestrian, mengurangi dampak kemacetan arus lalu lintas, dan dapat memberikan efisiensi waktu sesuai dengan kondisi serta jumlah pedestrian. Berdasarkan hasil pengujian, sistem ini dapat menghitung jumlah pedestrian dengan kecepatan frame rate 26.42 fps dan dapat mendeteksi pedestrian dalam beberapa skenario pergerakan dengan luasan jangkauan kamera hingga 20.88 m2 pada tinggi 3.25 meter dan sudut 45 derajat.
=======================================================================================================================================
Walking is a fast way to complete a short trip in an urban area. Various supporting facilities for pedestrians, one of which is the crossing facility infrastructure. In Indonesia, based on data from the National Police in the last 6 months, accidents on pedestrians were 8309 people, and 379 people were involved in crossing zebra crossing facilities [3]. The use of these facilities is considered to be less than optimal for pedestrians to cross which until now has been a fixed time, so that pedestrian users who cross in a hurry are higher. These obstacles can be overcome by the technology of deep learning object detection to determine pedestrian crossing times based on the analysis of the number of pedestrians and the detection of pedestrians using pelican-crossing by capturing images or videos from the IP camera. The existence of an adaptive crossing time arrangement system is expected to reduce the occurrence of negligence and increase security and comfort for pedestrians, reduce the impact of traffic congestion, and can provide time efficiency in accordance with the conditions and number of pedestrians. Based on the test results, this system can calculate the number of pedestrians with the speed of frame rate 26.42 fps and can facilitate pedestrians in various scenarios with coverage of cameras up to 20.88 m2 at high 3.25 meters and a camera angle of 45 degrees.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Pedestrian, Convolutional Neural Network, IP Camera , Object Detection
Subjects:	Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science) T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK5105.546 Computer algorithms T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK5105.585 TCP/IP (Computer network protocol) T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK7882.P3 Pattern recognition systems
Divisions:	Faculty of Electrical Technology > Computer Engineering > 90243-(S1) Undergraduate Thesis
Depositing User:	randy Putra Resha
Date Deposited:	06 Aug 2026 04:17
Last Modified:	06 Aug 2026 04:17
URI:	http://repository.its.ac.id/id/eprint/70465

Actions (login required)

View Item