Szczegóły publikacji
Opis bibliograficzny
Application of open deep learning datasets for building vision-based bridge condition monitoring solutions / K. TOMASZKIEWICZ, T. OWERKO // W: Bridge maintenance, safety, management, digitalization and sustainability : proceedings of the 12th international conference on bridge maintenance, safety and management, 24-28 June 2024, Copenhagen, Denmark / eds. Jens Sandager Jensen, Dan M. Frangopol, Jacob Wittrup Schmidt. — London : CRC Press, 2024. — Dod. ISBN: 978-1-032-77560-9 (pbk). — ISBN: 978-1-032-77040-6; e-ISBN: 978-1-003-48375-5. — S. 1496–1503. — Bibliogr. s. 1502–1503, Abstr.
Autorzy (2)
Dane bibliometryczne
| ID BaDAP | 154344 |
|---|---|
| Data dodania do BaDAP | 2024-07-11 |
| Tekst źródłowy | URL |
| DOI | 10.1201/9781003483755-175 |
| Rok publikacji | 2024 |
| Typ publikacji | materiały konferencyjne (aut.) |
| Otwarty dostęp |  |
| Creative Commons | |
| Wydawca | CRC Press / Balkema |
Abstract
The operational phase of bridges is the longest phase in the entire structure’s life cycle. During this phase, defects occurring in earlier stages of the life cycle may become apparent. Structural condition monitoring and defect detection at an early stage of their formation make it possible to react faster, protect the damage and stop or delay the degradation process. To support the engineer’s work, deep learning-based solutions are becoming popular. However, to make solutions based on such algorithms effective and helpful for the engineer, it is necessary to provide a good dataset to train them. This paper summarizes publicly available open datasets of bridge damage images that can be used to train DL algorithms for classification, object detection and segmentation. The presented approach refers to the usability of datasets from the point of view of training DL algorithms, as well as the usability of algorithms trained on their basis to support the bridge engineer’s work.
