PERANCANGAN DAN IMPLEMENTASI MULTI-TASK CNN SEBAGAI SISTEM PENDETEKSI WAJAH BERBASIS MIKROKONTROLER ESP32

LEANDITA, REGINA and Ubaya, Huda (2020) PERANCANGAN DAN IMPLEMENTASI MULTI-TASK CNN SEBAGAI SISTEM PENDETEKSI WAJAH BERBASIS MIKROKONTROLER ESP32. Undergraduate thesis, Sriwijaya University.

	Text RAMA_56201_09011381520070.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (9MB) | Request a copy
	Text RAMA_56201_09011381520070_TURNITIN.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (11MB) | Request a copy
Preview	Text RAMA_56201_09011381520070_0216068101_01_front_ref.pdf - Accepted Version Available under License Creative Commons Public Domain Dedication. Download (3MB) | Preview
	Text RAMA_56201_09011381520070_0216068101_02.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (843kB) | Request a copy
	Text RAMA_56201_09011381520070_0216068101_03.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (414kB) | Request a copy
	Text RAMA_56201_09011381520070_0216068101_04.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (2MB) | Request a copy
	Text RAMA_56201_09011381520070_0216068101_05.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (88kB) | Request a copy
	Text RAMA_56201_09011381520070_0216068101_06_ref.pdf - Bibliography Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (401kB) | Request a copy
	Text RAMA_56201_09011381520070_0216068101_07_lamp.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (1MB) | Request a copy

Abstract

One of the most common problems in the process of CNN is requires large computational resources. High computing makes CNN unable to be implemented on lighter devices such as microcontrollers as mobile devices. In this study, we aim to implement the Multi-Task CNN that works cascade that divides the computational load of machine learning is done on training. As the application to mobile devices, use microcontroller ESP32. From experiments using a sample of photos with different 5-facing angle, masks and without masks respectively 5 photos, showed ESP32 can detect faces 18 of 25 face images. Research also continues by comparing Multi-Task CNN with Haar-Cascade method. With a sample of similar photos, obtained accuracy 90% for Multi-Task CNN and 60% for Haar-Cascade. Other final results are the average CNN Multi-Task testing process time of 3.29 s for sample without masker sample and for sample using masker. Meanwhile, Haar-Cascade 0.21 s for sample without masker sample and for sample using masker.

Item Type:	Thesis (Undergraduate)
Uncontrolled Keywords:	CNN, Image Processing
Subjects:	Q Science > Q Science (General) > Q334-342 Computer science. Artificial intelligence. Algorithms. Robotics. Automation. T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK7885-7895 Computer engineering. Computer hardware
Divisions:	09-Faculty of Computer Science > 56201-Computer Systems (S1)
Depositing User:	Users 6796 not found.
Date Deposited:	13 Aug 2020 04:42
Last Modified:	13 Aug 2020 04:42
URI:	http://repository.unsri.ac.id/id/eprint/32986

Actions (login required)

View Item