diplomsko delo
Jan Banko (Author), Božidar Potočnik (Mentor), Martin Šavc (Co-mentor)

Abstract

V diplomskem delu se ukvarjamo s prepoznavo jedi iz digitalnih slik s pomočjo konvolucijskih nevronskih mrež. Namen diplomskega dela je razvoj in implementacija sistema, ki je zmožen prepoznati hrano na digitalni sliki. Natančneje smo preučili delovanje konvolucijskih nevronskih mrež ter postopek prepoznavanja objektov. Opisali smo tudi uporabljene algoritme za detekcijo objektov, ki uporabljajo konvolucijske nevronske mreže. Pri implementaciji razpoznavalnika hrane smo se omejili na 8 različnih kategorij hrane. Pri testiranju na podatkovni zbirki »The Food-101 Data Set« je na množici 2400 slik najboljši izmed uporabljenih modelov detektorjev dosegel natančnost prepoznavanja 95,59 % pri uporabi metrike »PASCAL VOC 2010« ter 72,1 % pri uporabi metrike »COCO«.

Keywords

računalniški vid;prepoznavanje hrane;konvolucijske nevronske mreže;diplomske naloge;

Data

Language: Slovenian
Year of publishing:
Typology: 2.11 - Undergraduate Thesis
Organization: UM FERI - Faculty of Electrical Engineering and Computer Science
Publisher: J. Banko
UDC: 004.93:004.032.26(043.2)
COBISS: 21768982 Link will open in a new window
Views: 1712
Downloads: 268
Average score: 0 (0 votes)
Metadata: JSON JSON-RDF JSON-LD TURTLE N-TRIPLES XML RDFA MICRODATA DC-XML DC-RDF RDF

Other data

Secondary language: English
Secondary title: Food recognition from digital images using convolutional neural networks
Secondary abstract: In the thesis we are dealing with food recognition from digital images using convolutional neural networks. The purpose of this thesis is to develop and implement a system, capable of detecting food items in a digital image. We thoroughly studied how convolutional neural networks work, and examined the object detection pipeline. We describe convolutional neural network based object detection algorithms that were used in the thesis. In the implementation of the food detection system we limited ourselves to 8 different food categories. During testing on »The Food-101 Data Set« using 2400 images, the best object detection model of those that were used achieved a classification rate of 95.59 % when using the »PASCAL VOC 2010« metric and 72.1 % when using the »COCO« metric.
Secondary keywords: computer vision;food recognition;convolutional neural networks;
URN: URN:SI:UM:
Type (COBISS): Bachelor thesis/paper
Thesis comment: Univ. v Mariboru, Fak. za elektrotehniko, računalništvo in informatiko, Računalništvo in informacijske tehnologije
Pages: VII, 48 f.
ID: 10950988