Niemeijer, Joshua und Srinivas, Gurucharan und Leich, Andreas und Battistella, Federico (2023) An Approach for Fusing Two Training-Datasets with Partially Overlapping Classes. In: 17th IEEE International Conference on Semantic Computing, ICSC 2023. 17th International Conference on Semantic Computing (ICSC), 2023-02-01 - 2023-02-03, Laguna Hills, CA, USA. ISBN 978-166548263-9.
PDF
- Nur DLR-intern zugänglich
2MB |
Kurzfassung
Supervised deep learning techniques in image processing require training data, typically consisting of manually labeled ground truth annotations. Since manual labeling is costly, using as many existing training datasets as possible is necessary. This paper introduces a novel approach for combining training datasets into a new one. The naive approach to this is a plain concatenation of the existing datasets. However, this approach fails with partially overlapping datasets when certain annotated instances specific to one dataset also appear in the other dataset without their annotations. Therefore, we present a novel method for combining existing training datasets using a pseudo-labeling technique with uncertainty quantification. The effectiveness of our method is evaluated by fusing two datasets consisting of partially overlapping traffic sign annotations in street view images. The results demonstrate that the pseudo-labeling errors weigh less than those introduced by the naive fusion. Furthermore, our work provides evidence for practitioners to use a pseudolabeling-based fusion technique with uncertainty quantificationrather than naively combining training datasets into a new one.
elib-URL des Eintrags: | https://elib.dlr.de/198542/ | ||||||||||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Dokumentart: | Konferenzbeitrag (Vortrag) | ||||||||||||||||||||
Titel: | An Approach for Fusing Two Training-Datasets with Partially Overlapping Classes | ||||||||||||||||||||
Autoren: |
| ||||||||||||||||||||
Datum: | 2023 | ||||||||||||||||||||
Erschienen in: | 17th IEEE International Conference on Semantic Computing, ICSC 2023 | ||||||||||||||||||||
Referierte Publikation: | Ja | ||||||||||||||||||||
Open Access: | Nein | ||||||||||||||||||||
Gold Open Access: | Nein | ||||||||||||||||||||
In SCOPUS: | Ja | ||||||||||||||||||||
In ISI Web of Science: | Ja | ||||||||||||||||||||
ISBN: | 978-166548263-9 | ||||||||||||||||||||
Status: | veröffentlicht | ||||||||||||||||||||
Stichwörter: | Dataset Fusion, Traffic Sign detection, Deep Learning, Uncertainty Quantification, Pseudo-Labeling | ||||||||||||||||||||
Veranstaltungstitel: | 17th International Conference on Semantic Computing (ICSC) | ||||||||||||||||||||
Veranstaltungsort: | Laguna Hills, CA, USA | ||||||||||||||||||||
Veranstaltungsart: | internationale Konferenz | ||||||||||||||||||||
Veranstaltungsbeginn: | 1 Februar 2023 | ||||||||||||||||||||
Veranstaltungsende: | 3 Februar 2023 | ||||||||||||||||||||
Veranstalter : | IEEE | ||||||||||||||||||||
HGF - Forschungsbereich: | Luftfahrt, Raumfahrt und Verkehr | ||||||||||||||||||||
HGF - Programm: | Verkehr | ||||||||||||||||||||
HGF - Programmthema: | Straßenverkehr | ||||||||||||||||||||
DLR - Schwerpunkt: | Verkehr | ||||||||||||||||||||
DLR - Forschungsgebiet: | V ST Straßenverkehr | ||||||||||||||||||||
DLR - Teilgebiet (Projekt, Vorhaben): | V - KoKoVI - Koordinierter kooperativer Verkehr mit verteilter, lernender Intelligenz | ||||||||||||||||||||
Standort: | Berlin-Adlershof , Braunschweig | ||||||||||||||||||||
Institute & Einrichtungen: | Institut für Verkehrssystemtechnik Institut für Verkehrssystemtechnik > Kooperative Systeme, BS Institut für Verkehrssystemtechnik > Informationsgewinnung und Modellierung, BA | ||||||||||||||||||||
Hinterlegt von: | Niemeijer, Joshua | ||||||||||||||||||||
Hinterlegt am: | 08 Dez 2023 14:43 | ||||||||||||||||||||
Letzte Änderung: | 24 Apr 2024 20:58 |
Nur für Mitarbeiter des Archivs: Kontrollseite des Eintrags