Habibi, Mochammad Reza (2021) Peningkatan Performansi Single Shot Detector Dan Modifikasi Kalibrasi Zhang Untuk Estimasi Jarak Dalam Mendukung Social Distancing. Masters thesis, Institut Teknologi Sepuluh Nopember.
Text
06111950010004-Master_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2023. Download (4MB) |
Abstract
Coronavirus Disease (COVID19) telah memberikan dampak yang luar biasa pada seluruh dunia tak-terkecuali di Indonesia. Rata – rata tingkat penularan COVID19 (R-naught) adalah 2-3 orang yang masih tergolong tinggi. Penerapan protokol kesehatan seperti social distancing adalah salah satu mekanisme yang dapat mengurangi tingkat penularan. Sistem penerapan social distancing dapat dilakukan secara manual tetapi membutuhkan sumber daya yang besar. Maka pada penelitian ini dikembangkan Single Shot Detector yang telah ditingkatkan performanya menggunakan Mobilenet dan Kalibrasi Zhang yang telah dimodifikasi untuk kamera monocular dalam mendukung social distancing secara otomatis. Metode kalibrasi berbeda yang berbasis sifat geometris objek juga diterapkan dan dikomparasi untuk mendapatkan hasil terbaik. Dari hasil ujicoba, didapat performa kecepatan untuk deteksi objek dari Single Shot Detector meningkat 82% menggunakan Mobilenet. Sedangkan dalam mengestimasi jarak didapat rerata tingkat error 5% menggunakan Kalibrasi Zhang dan 26% menggunakan kalibrasi berbasis sifat geometris objek.
==================================================================================================
The Coronavirus Disease (COVID19) has brought a terrific crisis globally. The transmission rate (R-naught) is high to wit an infected person may transmit to 2-3 people on average. To maintain the health protocol such as social distancing is a proven mechanism to reduce COVID19 transmission rate. It is possible to do it manually however there are drawbacks such as expensive resource and excessive cost. To overcome the aforementioned difficulties, a system utilizes Single Shot Detector which has been improved using Mobilenet and Zhang Calibration which has been modified for monocular camera have been developed in support of social distancing automatically. Furthermore, different camera calibration method based on the geometric properties of the object have also been applied and compared. From the obtained result, speed performance of Single Shot Detector for object detection increased by 82% using Mobilenet. Whereas, in estimating the distance, Zhang Calibration achieved the average error rate of 5% while the other camera calibration method achieved the average error rate of 26%.
Item Type: | Thesis (Masters) |
---|---|
Uncontrolled Keywords: | Mobilenet Single Shot Detector, Estimasi Jarak, Kalibrasi Kamera, Social Distancing, Kamera Monocular, Distance Estimation, Camera Calibration, Monocular Camera. |
Subjects: | Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines. Q Science > QA Mathematics > QA336 Artificial Intelligence Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science) Q Science > QC Physics > QC271.8.C3 Calibration R Medicine > R Medicine (General) > R858 Deep Learning R Medicine > RA Public aspects of medicine > RA644.C67 COVID-19 (Disease) 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: | Mochammad Reza Habibi |
Date Deposited: | 28 Aug 2021 15:01 |
Last Modified: | 28 Aug 2021 15:01 |
URI: | http://repository.its.ac.id/id/eprint/91098 |
Actions (login required)
View Item |