Traffic Light Recognition for Real Scenes Based on Image Processing and Deep Learning
| Autor: | Xinliang Cao, Zhenhua Chao, Mingliang Che, Mingjun Che |
|---|---|
| Jazyk: | angličtina |
| Rok vydání: | 2020 |
| Předmět: |
fractal dimension
Computer science business.industry Traffic light recognition Deep learning Perspective (graphical) General Engineering perspective relationship 68U10 Image processing Fractal dimension Traffic signal color features Computer vision Artificial intelligence SqueezeNet Architecture business Reference model Mobile device |
| Zdroj: | COMPUTING AND INFORMATICS; Vol. 39 No. 3 (2020): Computing and Informatics; 439-463 |
| ISSN: | 1335-9150 2585-8807 |
| Popis: | Traffic light recognition in urban environments is crucial for vehicle control. Many studies have been devoted to recognizing traffic lights. However, existing recognition methods still face many challenges in terms of accuracy, runtime and size. This paper presents a novel robust traffic light recognition approach that takes into account these three aspects based on image processing and deep learning. The proposed approach adopts a two-stage architecture, first performing detection and then classification. In the detection, the perspective relationship and the fractal dimension are both considered to dramatically reduce the number of invalid candidate boxes, i.e. region proposals. In the classification, the candidate boxes are classified by SqueezeNet. Finally, the recognized traffic light boxes are reshaped by postprocessing. Compared with several reference models, this approach is significantly competitive in terms of accuracy and runtime. We show that our approach is lightweight, easy to implement, and applicable to smart terminals, mobile devices or embedded devices in practice. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|