Traffic Light Recognition Based on Binary Semantic Segmentation Network
A traffic light recognition system is a very important building block in an advanced driving assistance system and an autonomous vehicle system. In this paper, we propose a two-staged deep-learning-based traffic light recognition method that consists of a pixel-wise semantic segmentation technique a...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2019-04-01
|
| Series: | Sensors |
| Subjects: | |
| Online Access: | https://www.mdpi.com/1424-8220/19/7/1700 |
| _version_ | 1798038501327896576 |
|---|---|
| author | Hyun-Koo Kim Kook-Yeol Yoo Ju H. Park Ho-Youl Jung |
| author_facet | Hyun-Koo Kim Kook-Yeol Yoo Ju H. Park Ho-Youl Jung |
| author_sort | Hyun-Koo Kim |
| collection | DOAJ |
| description | A traffic light recognition system is a very important building block in an advanced driving assistance system and an autonomous vehicle system. In this paper, we propose a two-staged deep-learning-based traffic light recognition method that consists of a pixel-wise semantic segmentation technique and a novel fully convolutional network. For candidate detection, we employ a binary-semantic segmentation network that is suitable for detecting small objects such as traffic lights. Connected components labeling with an eight-connected neighborhood is applied to obtain bounding boxes of candidate regions, instead of the computationally demanding region proposal and regression processes of conventional methods. A fully convolutional network including a convolution layer with three filters of (1 × 1) at the beginning is designed and implemented for traffic light classification, as traffic lights have only a set number of colors. The simulation results show that the proposed traffic light recognition method outperforms the conventional two-staged object detection method in terms of recognition performance, and remarkably reduces the computational complexity and hardware requirements. This framework can be a useful network design guideline for the detection and recognition of small objects, including traffic lights. |
| first_indexed | 2024-04-11T21:41:05Z |
| format | Article |
| id | doaj.art-4e4643e93dc540dda4799ece30e44256 |
| institution | Directory Open Access Journal |
| issn | 1424-8220 |
| language | English |
| last_indexed | 2024-04-11T21:41:05Z |
| publishDate | 2019-04-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Sensors |
| spelling | doaj.art-4e4643e93dc540dda4799ece30e442562022-12-22T04:01:35ZengMDPI AGSensors1424-82202019-04-01197170010.3390/s19071700s19071700Traffic Light Recognition Based on Binary Semantic Segmentation NetworkHyun-Koo Kim0Kook-Yeol Yoo1Ju H. Park2Ho-Youl Jung3Department of Information and Communication Engineering, Yeungnam University, Gyeongsan 38544, KoreaDepartment of Information and Communication Engineering, Yeungnam University, Gyeongsan 38544, KoreaDepartment of Electrical Engineering, Yeungnam University, Gyeongsan 38544, KoreaDepartment of Information and Communication Engineering, Yeungnam University, Gyeongsan 38544, KoreaA traffic light recognition system is a very important building block in an advanced driving assistance system and an autonomous vehicle system. In this paper, we propose a two-staged deep-learning-based traffic light recognition method that consists of a pixel-wise semantic segmentation technique and a novel fully convolutional network. For candidate detection, we employ a binary-semantic segmentation network that is suitable for detecting small objects such as traffic lights. Connected components labeling with an eight-connected neighborhood is applied to obtain bounding boxes of candidate regions, instead of the computationally demanding region proposal and regression processes of conventional methods. A fully convolutional network including a convolution layer with three filters of (1 × 1) at the beginning is designed and implemented for traffic light classification, as traffic lights have only a set number of colors. The simulation results show that the proposed traffic light recognition method outperforms the conventional two-staged object detection method in terms of recognition performance, and remarkably reduces the computational complexity and hardware requirements. This framework can be a useful network design guideline for the detection and recognition of small objects, including traffic lights.https://www.mdpi.com/1424-8220/19/7/1700advanced driver assistance systemartificial neural networksbinary semantic segmentationdeep learningtraffic light detectiontraffic light recognition |
| spellingShingle | Hyun-Koo Kim Kook-Yeol Yoo Ju H. Park Ho-Youl Jung Traffic Light Recognition Based on Binary Semantic Segmentation Network Sensors advanced driver assistance system artificial neural networks binary semantic segmentation deep learning traffic light detection traffic light recognition |
| title | Traffic Light Recognition Based on Binary Semantic Segmentation Network |
| title_full | Traffic Light Recognition Based on Binary Semantic Segmentation Network |
| title_fullStr | Traffic Light Recognition Based on Binary Semantic Segmentation Network |
| title_full_unstemmed | Traffic Light Recognition Based on Binary Semantic Segmentation Network |
| title_short | Traffic Light Recognition Based on Binary Semantic Segmentation Network |
| title_sort | traffic light recognition based on binary semantic segmentation network |
| topic | advanced driver assistance system artificial neural networks binary semantic segmentation deep learning traffic light detection traffic light recognition |
| url | https://www.mdpi.com/1424-8220/19/7/1700 |
| work_keys_str_mv | AT hyunkookim trafficlightrecognitionbasedonbinarysemanticsegmentationnetwork AT kookyeolyoo trafficlightrecognitionbasedonbinarysemanticsegmentationnetwork AT juhpark trafficlightrecognitionbasedonbinarysemanticsegmentationnetwork AT hoyouljung trafficlightrecognitionbasedonbinarysemanticsegmentationnetwork |