prikaz prve stranice dokumenta Identifikacija objekata u slici dobivenoj s kamere u vozilu uz korištenje ROS-a
  Download
PDF 3.46 MB
master's thesis
Identifikacija objekata u slici dobivenoj s kamere u vozilu uz korištenje ROS-a
2019. urn:nbn:hr:200:675935
Jelić, Borna
Josip Juraj Strossmayer University of Osijek
Faculty of Electrical Engineering, Computer Science and Information Technology Osijek
Department of Communications
Chair of Electronics and Microelectronics

Cite this document

Jelić, B. (2019). Identifikacija objekata u slici dobivenoj s kamere u vozilu uz korištenje ROS-a (Master's thesis). Retrieved from https://urn.nsk.hr/urn:nbn:hr:200:675935

Jelić, Borna. "Identifikacija objekata u slici dobivenoj s kamere u vozilu uz korištenje ROS-a." Master's thesis, Josip Juraj Strossmayer University of Osijek, Faculty of Electrical Engineering, Computer Science and Information Technology Osijek, 2019. https://urn.nsk.hr/urn:nbn:hr:200:675935

Jelić, Borna. "Identifikacija objekata u slici dobivenoj s kamere u vozilu uz korištenje ROS-a." Master's thesis, Josip Juraj Strossmayer University of Osijek, Faculty of Electrical Engineering, Computer Science and Information Technology Osijek, 2019. https://urn.nsk.hr/urn:nbn:hr:200:675935

Jelić, B. (2019). 'Identifikacija objekata u slici dobivenoj s kamere u vozilu uz korištenje ROS-a', Master's thesis, Josip Juraj Strossmayer University of Osijek, Faculty of Electrical Engineering, Computer Science and Information Technology Osijek, accessed 21 March 2023, https://urn.nsk.hr/urn:nbn:hr:200:675935

Jelić B. Identifikacija objekata u slici dobivenoj s kamere u vozilu uz korištenje ROS-a [Master's thesis]. Osijek: Josip Juraj Strossmayer University of Osijek, Faculty of Electrical Engineering, Computer Science and Information Technology Osijek; 2019 [cited 2023 March 21] Available at: https://urn.nsk.hr/urn:nbn:hr:200:675935

B. Jelić, "Identifikacija objekata u slici dobivenoj s kamere u vozilu uz korištenje ROS-a", Master's thesis, Josip Juraj Strossmayer University of Osijek, Faculty of Electrical Engineering, Computer Science and Information Technology Osijek, Osijek, 2019. Available at: https://urn.nsk.hr/urn:nbn:hr:200:675935

Please login to the repository to save this object to your list.
Metadata
TitleIdentifikacija objekata u slici dobivenoj s kamere u vozilu uz korištenje ROS-a
Title (english)Identifying objects in image captured from the vehicle camera using ROS
AuthorBorna Jelić
MentorMarijan Herceg (mentor)
Committee memberMarijan Herceg (predsjednik povjerenstva)
Committee memberMario Vranješ (član povjerenstva)
Committee memberRatko Grbić (član povjerenstva)
GranterJosip Juraj Strossmayer University of Osijek
Faculty of Electrical Engineering, Computer Science and Information Technology Osijek
(Department of Communications)
(Chair of Electronics and Microelectronics)
Osijek
Defense date and country2019-09-18, Croatia
Scientific / art field,
discipline and subdiscipline		TECHNICAL SCIENCES
Computing
Artificial Intelligence
Abstract
U radu su opisane određene postojeće metode detekcije objekata u slikama te je dan pregled korištenja istih za primjenu u autonomnim vozilima. Jedna od najvažnijih funkcija koju autonomna vozila trebaju obavljati je pravovremena i precizna detekcija objekata i predstavlja ozbiljan problem za inženjere koji se time bave. U sklopu ovog rada izrađena su tri rješenja za identifikaciju objekata u slici dobivenoj s prednje kamere na vozilu: jedno zasnovano na Viola-Jones algoritmu i dva zasnovana na YOLOv3 i Tiny YOLOv3 konvolucijskoj neuronskoj mreži. Objekti koji se mogu detektirati su vozila, pješaci, biciklisti, prometni znakovi podijeljeni u više kategorija i semafori. Vlastiti algoritmi izrađeni su u C++ programskom jeziku unutar ROS radnog okruženja te je provedeno njihovo testiranje na danim skupovima slika i video sekvenci. Rezultati su pokazali da algoritam zasnovan na YOLOv3 postiže znatno bolje rezultate od onog zasnovanog na Viola-Jones algoritmu. Međutim, i on kao takav ima prostora za doradu koja bi mu povećala razinu performansi s obzirom na preciznost, točnost i brzinu inferencije.
Abstract (english)
This master’s thesis describes certain existing object detection methods in images and gives an overview of their usage in autonomous vehicles. One of the most important functions autonomous vehicles must perform is timely and precise object detection, which presents a serious problem for engineers in the field. As a part of this master’s thesis, three solutions were developed regarding object detection in images from the vehicle camera, i.e., one based on the Viola-Jones algorithm and the other two based on YOLOv3 and Tiny YOLOv3 convolutional neural networks. Objects that could be detected are vehicles, pedestrians, cyclists, traffic signs divided into multiple categories and traffic lights. The algorithms were made in C++ programming language using ROS framework and tested on a set of test images and video sequences. The results show that the YOLOv3 based algorithm achieves much better results than the Viola-Jones one. However, it too has space for improvements in terms of its performances regarding precision, accuracy and inference speed.
Keywords
Keywords (english)
Languagecroatian
URN:NBNurn:nbn:hr:200:675935
Study programmeTitle: Graduate University Study Programme in Computer Engineering
Study programme type: university
Study level: graduate
Academic / professional title: magistar/magistra inženjer/inženjerka računarstva (magistar/magistra inženjer/inženjerka računarstva)
Type of resourceText
File originBorn digital
Access conditionsOpen access
Terms of useLicence In copyright icon
Created on2019-12-24 06:42:07