Transformers Neural Networks Applications in Different Computer Vision Tasks
Data
2023
Tytuł czasopisma
ISSN czasopisma
Tytuł tomu
Wydawca
Wydawnictwo Politechniki Łódzkiej
Lodz University of Technology Press
Lodz University of Technology Press
Abstrakt
Transformers architectures are one of the latest inventions in the
field of deep learning. Originally dedicated to NLP, they begin to find use
in computer vision too. In this paper, we briefly describe the idea behind
vision transformers and present a few examples, where we utilised them in
our research, focusing on the field of medical images and autonomous driving.
We show, that vision transformers can be used in various tasks, such as
detection or classification, as well as explain how some of their drawbacks
can be mitigated with a transfer learning approach.
Opis
Słowa kluczowe
transformers, neural networks, computer vision, classification, detection, segmentation, transformatory, sieci neuronowe, wizja komputerowa, klasyfikacja, detekcja, segmentacja
Cytowanie
Brodzicki A., Piekarski M., Kostuch A., Noworolnik F., Aleksandrowicz M., Wójcicka A., Jaworek-Korjakowska J., Transformers Neural Networks Applications in Different Computer Vision Tasks. W: Progress in Polish Artificial Intelligence Research 4, Wojciechowski A. (Ed.), Lipiński P. (Ed.)., Seria: Monografie Politechniki Łódzkiej Nr. 2437, Wydawnictwo Politechniki Łódzkiej, Łódź 2023, s. 73-79, ISBN 978-83-66741-92-8, doi: 10.34658/9788366741928.10.