Round Traffic Sign Detection Algorithm

Traffic signs are an important part of autonomous driving and intelligent transportation. It provides instructions for pedestrians and vehicles and is critical to road traffic safety. However, existing detection algorithms cannot achieve real-time high-precision detection. Therefore, this paper prop...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:Journal of physics. Conference series Ročník 2179; číslo 1; s. 12034 - 12039
Hlavní autoři: Huang, Yinrong, Wang, Bing, Yuan, Xiemin
Médium: Journal Article
Jazyk:angličtina
Vydáno: Bristol IOP Publishing 01.01.2022
Témata:
Algorithms
Artificial neural networks
Embedded systems
Hough transformation
Machine learning
Pedestrians
Physics
Real time
Traffic control
Traffic signs
ISSN:1742-6588, 1742-6596
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:Traffic signs are an important part of autonomous driving and intelligent transportation. It provides instructions for pedestrians and vehicles and is critical to road traffic safety. However, existing detection algorithms cannot achieve real-time high-precision detection. Therefore, this paper proposes an algorithm that combines traditional methods with deep learning to detect circular traffic signs. Based on the HSV color space, the red and blue channel images are separated, and the candidate regions of the original image are extracted using the Hough transform. The shallow convolutional neural network (CNN) classifier rejects not traffic signs and classifies traffic signs. Experiments show that the algorithm is real and effective. On the CPU platform, the average accuracy rate is 96.2%, and the detection speed reaches 0.3 s / frame. Under the condition of ensuring the average accuracy rate, the detection speed is greatly reduced. The algorithm achieves the fastest speed, which makes real-time high-precision detection possible. The algorithm is more suitable for vehicle embedded systems.
Bibliografie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:1742-6588
1742-6596
DOI:10.1088/1742-6596/2179/1/012034