Faculty of Information Technology, BUT

Publication Details

Vehicle Fine-grained Recognition Based on Convolutional Neural Networks for Real-world Applications

ŠPAŇHEL Jakub, SOCHOR Jakub and MAKAROV Aleksej. Vehicle Fine-grained Recognition Based on Convolutional Neural Networks for Real-world Applications. In: 2018 14th Symposium on Neural Networks and Applications (NEUREL). Belgrade: IEEE Signal Processing Society, 2018, pp. 1-5. ISBN 978-1-5386-6974-7.
Czech title
Rozpoznání modelů vozidel založené na neuronových sítí pro aplikaci v reálných podmínkách
Type
conference paper
Language
english
Authors
Špaňhel Jakub, Ing. (DCGM FIT BUT)
Sochor Jakub, Ing. (DCGM FIT BUT)
Makarov Aleksej (Vlatacom d.o.o.)
Keywords
convolutional neural networks, similar vehicle type search, vehicle fine-grained recognition, vehicle reidentification  
Abstract
We explore the implementation of vehicle
fine-grained type and color recognition based on neural
networks in a real-world application. We suggest changes to
the previously published method with respect to capabilities
of low-powered devices, such as Nvidia Jetson. Experimental
evaluation shows that the accuracy of MobileNet net slightly
decreases compared to ResNet-50 from 89.55% to 86.13%
while inference is 2.4× faster on Jetson.
Published
2018
Pages
1-5
Proceedings
2018 14th Symposium on Neural Networks and Applications (NEUREL)
Conference
2018 14th Symposium on Neural Networks and Applications (NEUREL), SAVA Center Milentija Popovića 9 11070, Belgrade, Serbia, RS
ISBN
978-1-5386-6974-7
Publisher
IEEE Signal Processing Society
Place
Belgrade, RS
DOI
BibTeX
@INPROCEEDINGS{FITPUB11851,
   author = "Jakub \v{S}pa\v{n}hel and Jakub Sochor and Aleksej Makarov",
   title = "Vehicle Fine-grained Recognition Based on Convolutional Neural Networks for Real-world Applications",
   pages = "1--5",
   booktitle = "2018 14th Symposium on Neural Networks and Applications (NEUREL)",
   year = 2018,
   location = "Belgrade, RS",
   publisher = "IEEE Signal Processing Society",
   ISBN = "978-1-5386-6974-7",
   doi = "10.1109/NEUREL.2018.8587012",
   language = "english",
   url = "https://www.fit.vut.cz/research/publication/11851"
}
Back to top