Logo des Repositoriums
 

Classifying figures and illustrations in electronics datasheets: A comparative evaluation of recent computer vision models on a custom collection of 4000 technical documents

Kurzanzeige
dc.contributor.authorPerakis, Lymperis
dc.contributor.authorBalling, Julian
dc.contributor.authorBinder, Frank
dc.contributor.authorHeyer, Gerhard
dc.contributor.authorKreupl, Franz
dc.contributor.editorKlein, Maike
dc.contributor.editorKrupka, Daniel
dc.contributor.editorWinter, Cornelia
dc.contributor.editorWohlgemuth, Volker
dc.date.accessioned2023-11-29T14:50:24Z
dc.date.available2023-11-29T14:50:24Z
dc.date.issued2023
dc.description.abstractWe report findings from a comparative evaluation of several recent object detection models applied to a domain-specific use case in technical document analysis and graphics recognition. More specifically, we apply models from the EfficientDet and YOLO model families to detect and classify figures in electronics datasheets according to a custom classification scheme. We identify YOLOv7-D6 as the most accurate model in our study and show that it can successfully solve this task. We highlight an iterative approach to figure annotation in document page images for creating a comprehensive and balanced custom dataset for our use case. In our experiments, the object detection models show impressive performance levels on par with state-of-the-art results from the literature and related studies.en
dc.identifier.doi10.18420/inf2023_186
dc.identifier.isbn978-3-88579-731-9
dc.identifier.pissn1617-5468
dc.identifier.urihttps://dl.gi.de/handle/20.500.12116/43114
dc.language.isoen
dc.publisherGesellschaft für Informatik e.V.
dc.relation.ispartofINFORMATIK 2023 - Designing Futures: Zukünfte gestalten
dc.relation.ispartofseriesLecture Notes in Informatics (LNI) - Proceedings, Volume P-337
dc.subjectComputer Vision
dc.subjectObject Detection
dc.subjectDocument Analysis
dc.subjectGraphics Recognition
dc.subjectElectronic Design Automation
dc.subjectMachine Learning
dc.titleClassifying figures and illustrations in electronics datasheets: A comparative evaluation of recent computer vision models on a custom collection of 4000 technical documentsen
dc.typeText/Conference Paper
gi.citation.endPage1848
gi.citation.publisherPlaceBonn
gi.citation.startPage1833
gi.conference.date26.-29. September 2023
gi.conference.locationBerlin
gi.conference.sessiontitleWirtschaft, Management Industrie - Künstliche Intelligenz für kleine und mittlere Unternehmen (KI-KMU 2023)

Dateien

Originalbündel
1 - 1 von 1
Vorschaubild nicht verfügbar
Name:
09_02_03_Perakis.pdf
Größe:
402.52 KB
Format:
Adobe Portable Document Format
Herunterladen

Sammlungen

P337 - INFORMATIK 2023 - Designing Futures: Zukünfte gestalten