Zaznavanje prisotnosti človeka na osnovi vgradnega računalnika s kamero

Rostohar, Matej (author), Vrabič, Rok (mentor)

Zaključna naloga obravnava poblem zaznave človeka z vgradnim računalnikom in kamero, pri tem uporabljamo konvolucijske nevronske mreže in model Posenet. Vzpostavili smo sistem s kamero, LED diodo in računalnikom Google Coral, ki je opremljen z EdgeTPU enoto, ki je posebej namenjena uporabi konvolucijskih mrež. Sistem deluje tako, da ko kamera zazna prisotnost človeka, LED dioda zasveti. Tehnologija omogoča avtonomnim robotom, da so z okolico povezani na novem nivoju.

• Language: Slovenian
• Keywords: konvolucijske nevronske mreže, kamere, človek, zaznavanje, posenet
• Work type: Final paper
• Typology: 2.11 - Undergraduate Thesis
• Organization: FS - Faculty of Mechanical Engineering
• Place of publishing: Ljubljana
• Publisher: [M. Rostohar]
• Year: 2019
• Number of pages: X, 28 f.
• PID: 20.500.12556/RUL-109839
• UDC: 004.8:004.38:621.397.42(043.2)
• COBISS.SI-ID: 16946971
• Publication date in RUL: 09.09.2019

ROSTOHAR, Matej, 2019, Zaznavanje prisotnosti človeka na osnovi vgradnega računalnika s kamero [online]. Bachelor’s thesis. Ljubljana : M. Rostohar. [Accessed 5 April 2025]. Retrieved from: https://repozitorij.uni-lj.si/IzpisGradiva.php?lang=eng&id=109839

Secondary language

• Language: English
• Title: Detecting the presence of humans using an embedded computer with a camera

Abstract

Thesis addresses the problem of human detection using a built-in computer and camera. Main software structure uses convolutional neural networks in combination with the Posenet model. We've assembled a system for human detection whose main components are: camera, LED diode and Google Coral computer, equipped with EdgeTPU module for fast and reliable usage of convolutional neural networks. The working principle of the system is as follows: when the camera detects a human being, the LED lights up. Technology allows autonomous robots to have a new method of extracting and interpreting information from the enviroment.

• Keywords: convolutional neural networks, cameras, human, detection, posenet