가중치 초기화 컨벌루션 뉴럴 네트워크를 통한 대용량 이미지 도로 검출 및 태깅 시스템
Big data Tagging System for Road Detection using Convolutional Neural Network with Weight Initialization
0
조회
0
다운로드
1
인용
0
참고문헌
키워드
초록
공중이미지로부터 정보를 확실하게 추출하여 실용적인 프로그램에 적용하는 것은 어려운 문제입니다. 이 정보추출 문제 중 구체적인 문제로서 자동 도로 검출이 있습니다. 이는 장애물로 인하여도로가 가려지거나, 그림자가 씌워지는 문제, 넓이의 변화가 다양한 점, 유사도로지역과 같은 문제가존재합니다. 오랜 기간 연구되어온 주제이나, 아직까지 상업단계에서 실제 도시이미지를 대상으로 한프로그램은 발표되지 않았습니다. 우리는 이 문제를 해결할 방법으로 수십 개의 가중치로 구성된 필터를 복합적으로 구성한 컨벌루션 뉴럴 네트워크를 제안합니다. 또한 이 필터의 초기화에 따라서 실질적으로 성능을 향상시킬 수 있음을 시연합니다.
Abstract
Extracting information from aerial images is a difficult problem with practical applications. A specific case of this is the task of automatically detecting roads, which is challengingbecause of occlusions, shadows, and a wide variety of road. Automatic or semi-automatic roaddetection system is not currently on the market and has not been shown to work reliably on largedataset of urban images. We propose a method of detecting roads using a convolutional neural networkwith many trainable weights and filter maps. Also, we demonstrate that predictive performancecan be improved by initializing the weights of the filter maps.
참고문헌 (17)
- Qian, Rongqiang, "Road surface traffic sign detection with hybrid region proposal and fast RCNN", Natural Computation, Fuzzy Systems and Knowledge Discovery (ICNC-FSKD), 2016 12th International Conference on. IEEE (2016)
- Baltsavias, E.P, "Object extraction and revision by image analysis using existing geodata and knowledge: current status and steps towards operational systems", ISPRS Journal of Photogrammetry and Remote Sensing, 58, pp.129~151 (2004)
- Mayer, H., "Object extraction in photogrammetric computer vision", ISPRS Journal of Photogrammetry and Remote Sensing, 63, pp.213~222 (2008)
- Hu, J., "Road Network Extraction and Intersection Detection From Aerial Images by Tracking Road Footprints", IEEE Transactions on Geoscience and Remote Sensing, 45, pp.4144~4157 (2007)
- Schmidt, Alena, "Forest point processes for the automatic extraction of networks in raster data", ISPRS Journal of Photogrammetry and Remote Sensing, 126, pp.38~55 (2017)
- Krizhevsky, Alex, "Imagenet classification with deep convolutional neural networks", Advances in neural information processing systems (2012)
- Le, Quoc V., "A simple way to initialize recurrent networks of rectified linear units", arXiv preprint arXiv:1504.00941, ~ (2015)
- Mena, J. B., "An automatic method for road extraction in rural and semi-urban areas starting from high resolution satellite imagery", Pattern Recognition Letters, 26, pp.1201~1220 (2005)
- Geman, D., "An active testing model fortracking roads in satellite images", IEEE Transactions on Pattern Analysis and Machine Intelligence, 18, pp.1~14 (1995)
- Bhangale, Ujwala, "Identification of informal settlement using Remote Sensing Images", Proceedings of the International Conference on Advances in Information Communication Technology & Computing. ACM (2016)
- Dollar, P., "Supervised learning of edges and object boundaries", CVPR ’06:Proceedings of the 2006 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, pp.1964~1971 (2006)
- Jain, V., "Advances in Neural Information Processing Systems 21", pp.769~776 (2009)
- Juneja, Mamta, "Performance evaluation of edge detection techniques for images in spatial domain", international journal of computer theory and Engineering, 1(5), pp.614~ (2009)
- Hinton, G, "Training products of experts by minimizing contrastive divergence", Neural Computation, 14, pp.1771~1800 (2002)
- Wiedemann, C., "Empirical Evaluation Techniques in Computer Vision", pp.172~187 (1998)
- 안길승, "신경망을 이용한 SNS상에서의 정보확산 예측모형: Digg 사례를 중심으로", 정보기술아키텍처 연구, 13(4), pp.609~616 (2016), 한국엔터프라이즈아키텍처학회
- 이종학, "영상정보 시스템을 위한 다차원 저장구조의 설계기법", 정보기술아키텍처 연구, 13(4), pp.639~650 (2016), 한국엔터프라이즈아키텍처학회