Title:	Deep Learning for the Detection of Car Flap States
Authors:	Guérand, Benoît Scheer, Fabian Demetgül, Mustafa Fleischer, Jürgen
Citation:	WSCG 2022: full papers proceedings: 30. International Conference in Central Europe on Computer Graphics, Visualization and Computer Vision, p. 142-151.
Issue Date:	2022
Publisher:	Václav Skala - UNION Agency
Document type:	conferenceObject
URI:	http://hdl.handle.net/11025/49588
ISBN:	978-80-86943-33-6
ISSN:	2464-4617
Keywords:	hluboké učení;Resnet50;RetinaNet;mezery ve dveřích;detekce objektů otevřených klapek auta;konvoluční neuronové sítě;CNN;případ průmyslového použití;výrobní linka
Keywords in different language:	deep learning;Resnet50;RetinaNet;door gaps;object detection of open car flaps;convolutional neural networks;CNN;industrial use case;production line
Abstract in different language:	In recent years, deep learning and object detection has continuously attracted more attention. Especially in the automotive world where many car manufacturers are currently investigating its possible applications. On production lines, even if processes are more and more automatized mistakes can happen and hinder the performance of an industrial plant. In this study, a method and application of object detection-based deep learning algorithm to detect open flaps on cars, like doors, trunk, hood etc. is examined. With this approach, the advantages of gap detection in cars on production lines, specifically the application of Resnet50 Convolutional Neural Networks (CNNs) and transfer learning in an industrial use case, are demonstrated. We show how the problem of detecting open flaps on cars is modeled in a way that a CNN can be applied to this new kind of application and present a detailed evaluation of the results and challenges. Finally, many suggestions are given for future applications of similar algorithms.
Rights:	© Václav Skala - UNION Agency
Appears in Collections:	WSCG 2022: Full Papers Proceedings

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.