열화 도로 영상에서 블랙 아이스 영역의 실시간 분할
Real-time Segmentation of Black Ice Region in Infrared Road Images
1
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
본 논문에서는 운전자한테 실시간으로 블랙 아이스 경고를 보내기 위해서 도로 영상에서 블랙 아이스 영역 분할을 위한 다중 척도 팽창 컨볼루션 특징 융합에 기반한 딥러닝 모델을 제안한다. 제안한 다중척도 팽창 컨볼루션 특징 융합 네트워크는 인코더 블록에 서로 다른 팽창 비율 컨볼루션을 병렬로 추가하고, 서로 다른 해상도 특징 맵에서 서로 다른 팽창 비율을 설정하고, 다중 단계 특징 정보가 함께 융합된다. 다중 척도 팽창 컨볼루션 특징 융합은 수용 영역을 확장함과 동시에 공간의 세부 정보를 잘 보존하고 팽창컨볼루션의 효과성을 높임으로써 기존 모델보다 성능을 향상시킨다. 실험 결과를 통해 본 논문 제안한네트워크 모델은 병렬 평창 컨볼루션 수가 증가함에 따라 성능이 향상되는 것을 알 수 있었다. 제안한방법의 mIoU 값은 96.46%로 U-Net, FCN, PSPNet, ENet, LinkNet 등 기존 네트워크보다 높았다. 그리고파라미터는 1,858K개로, 기존 LinkNet모델보다 6배로 축소하였다. Jetson Nano에서 실험 결과 보면, 제안한 방법의 FPS는 3.63로 실시간으로 블랙 아이스 영역을 실시간으로 분할 할 수 있었다.
Abstract
In this paper, we proposed a deep learning model based on multi-scale dilated convolution feature fusion for the segmentation of black ice region in road image to send black ice warning to drivers in real time. In the proposed multi-scale dilated convolution feature fusion network, different dilated ratio convolutions are connected in parallel in the encoder blocks, and different dilated ratios are used in different resolution feature maps, and multi-layer feature information are fused together. The multi-scale dilated convolution feature fusion improves the performance by diversifying and expending the receptive field of the network and by preserving detailed space information and enhancing the effectiveness of diated convolutions. The performance of the proposed network model was gradually improved with the increase of the number of dilated convolution branch. The mIoU value of the proposed method is 96.46%, which was higher than the existing networks such as U-Net, FCN, PSPNet, ENet, LinkNet. The parameter was 1,858K, which was 6 times smaller than the existing LinkNet model. From the experimental results of Jetson Nano, the FPS of the proposed method was 3.63, which can realize segmentation of black ice field in real time.
참고문헌 (23)
- 2021 Edition (Statistics for 2020) Traffic Accident Statistical Analysis Report, , Road Traffic Authority(KoROAD), ~
- 박근영, "겨울철 노면에 발생하는 어는 비와 블랙아이스의 기상학적 분석에 관한 사례 연구", 한국환경과학회지, 26(7), pp.827~836 (2017), 한국환경과학회[DOI]
- O. P. Ghim, "Mechanistic Interpretation of Braking Distance Specifications and Pavement Friction Requirements", Journal of the Transportation Research Board, 2155, pp.145~157 (2010)[DOI]
- "Korea Traffic Accident Analysis System"[URL]
- R. Omer, "An Automatic Image Recognition System For Winter Road Surface Condition Classification", International IEEE Conference on Intelligent Transportation Systems, pp.1375~1379 (2010)
- S. Y. Tan, "Road surface icing detecting system based on multiple sensor", Transducer and Microsystem Technologies, 35(1), pp.107~109 (2016)[DOI]
- Q. Li, "Design of Road Icing Detection System Based on Opencv+Python", Journal of Shaanxi University of Science & Technology(Natural Science Edition), 35(2), pp.158~164 (2017)[DOI]
- Y. E. Abdalla, "Black Ice detection system using Kinect", 2017 IEEE 30th Canadian Conference on Electrical and Computer Engineering (CCECE), pp.1~4 (2017)
- H. Lee, "Black ice Detection Using CNN for the Prevention of Accidents in Automated Vehicle", International Conference on Computational Science and Computational Intelligence (CSCI), pp.1189~1192 (2020)
- Y. C. Du, "Visibility Analysis for Freeway Based on Comparison of Ordinary and Infrared Images", Journal of Transportation Systems Engineering and Information Technology, 16(4), pp.73~78 (2016)[DOI]
- J. Long, "Fully convolutional networks for semantic segmentation", IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp.3431~3440 (2015)
- M. Everingham, "The PASCAL Visual Object Classes Challenge: A Retrospective", International Journal of Computer Vision, 111, pp.98~136 (2015)[DOI]
- O. Ronneberger, "U-Net: Convolutional Networks for Biomedical Image Segmentation", International Conference on Medical Image Computing and Computer-Assisted Intervention(MICCAI), pp.234~241 (2015)
- H. Zhao, "Pyramid Scene Parsing Network", 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp.6230~6239 (2017)
- L. -C. Chen, "DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs", IEEE Transactions on Pattern Analysis and Machine Intelligence, 40(4), pp.834~848 (2018)[DOI]
- K. He, "Deep Residual Learning for Image Recognition", 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp.770~778 (2015)
- L Chen, "Encoder-Decoder with Atrous Separable Convolution for Semantic Image Segmentation", European Conference on Computer Vision(ECCV), pp.883~851 (2018)
- F. Chollet, "Xception: Deep Learning with Depthwise Separable Convolutions", IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp.1800~1807 (2017)
- F. Yu, "Multi-Scale Context Aggregation by Dilated Convolutions", ICLR (2015)
- A. Paszke, "ENet: A Deep Neural Network Architecture for Real-Time Semantic Segmentation", ~ (2016)
- A. Chaurasia, "LinkNet: Exploiting encoder representations for efficient semantic segmentation", IEEE Visual Communications and Image Processing, pp.1~4 (2017)[DOI]
- 이옥걸, "MobileNetV2 기반의 개선된 Lightweight 모델을 이용한 열화 도로 영상에서의 블랙 아이스 인식", 한국정보통신학회논문지, 25(12), pp.1835~1845 (2021), 한국정보통신학회[DOI]