DOI QR코드

DOI QR Code

원격 영상에서 심층 잔차 밀집 기반의 초고해상도 기법을 이용한 차량 검출 알고리즘

Vehicle Detection Algorithm Using Super Resolution Based on Deep Residual Dense Block for Remote Sensing Images

  • 권오설 (창원대학교 전기전자제어공학부)
  • Oh-Seol Kwon (School of Electrical Electronics and Control Engineering, Changwon National University)
  • 투고 : 2022.12.05
  • 심사 : 2023.01.04
  • 발행 : 2023.01.30

초록

원거리에서 특정 영역의 물리적 특성 또는 상황에 대한 정보를 얻기 위해 원격 탐사 영상에 객체 검출 기법이 연구되고 있다. 이때 저해상도인 원격 영상은 정보의 손실로 인해 객체 검출의 정확도가 떨어지는 문제가 발생한다. 본 논문에서는 이러한 문제점을 해결하기 위해 초고해상도 기법과 객체 검출 방법을 하나의 네트워크로 구성하여 원격 영상에서 객체 검출의 성능을 높이는 방법을 제안한다. 제안한 방법은 심층 잔차 밀집 기반의 네트워크를 구성하여 저해상도 영상에서 객체의 특징을 복원하고자 하였다. 추가적으로 이를 객체 검출 단계인 YOLOv5와 하나의 네트워크로 구성함으로써 객체 검출의 성능을 향상시키고자 하였다. 제안한 방법은 저해상도 영상을 위해 VEDAI 데이터를 이용하였으며 차량 검출에서 VISIBLE 기준으로 mAP@0.5에 대해 81.38%까지 향상됨을 확인하였다.

Object detection techniques are increasingly used to obtain information on physical characteristics or situations of a specific area from remote images. The accuracy of object detection is decreased in remote sensing images with low resolution because the low resolution reduces the amount of detail that can be captured in an image. A single neural network is proposed to joint the super-resolution method and object detection method. The proposed method constructs a deep residual-based network to restore object features in low-resolution images. Moreover, the proposed method is used to improve the performance of object detection by jointing a single network with YOLOv5. The proposed method is experimentally tested using VEDAI data for low-resolution images. The results show that vehicle detection performance improved by 81.38% on mAP@0.5 for VISIBLE data.

키워드

과제정보

이 논문은 2021학년도 창원대학교 연구교수 연구비에 의하여 연구되었음.

참고문헌

  1. M. Thomas and M. Farid, "Automatic Car Counting Method for Unmanned Aerial Vehicle Image," IEEE Transactions on Geoscience and Remote Sensing, Vol. 52, No. 3, pp.1635-1647, Mar. 2014. doi: https://doi.org/10.1109/TGRS.2013.2253108
  2. K Liu. and G. Mattyus, "Fast Multi-class Vehicle Detection on Aerial Images," IEEE Geosci. Remote Sens. Lett., Vol. 12, No. 9, pp. 1938-1942, 2015. doi: https://doi.org/10.1109/LGRS.2015.2439517
  3. Z. Shengjie, L. Jinghong, T. Yang. Z. Yujia, and L. Chenglong, "Rapid Vehicle Detection in Aerial Images under the Complex Background of Dense Urban Areas," Remote sensing, Vol. 14, No. 9, pp.1-22, 2022. doi: https://doi.org/10.3390/rs14092088
  4. A. Saeed, K. Salman, and B. Nick, "A Deep Journey into Super-resolution: A Survey," ACM Comput. Surv. Vol. 53, pp. 1-21. 2020. doi: https://doi.org/10.48550/arXiv.1904.07523
  5. Z. Jiandan, L. Tao, Y. Guangle, "Robust Vehicle Detection in Aerial Images Based on Cascaded Convolutional Neural Networks," Sensors, Vol. 17, No. 12, pp.1-17, 2017 doi: https://doi.org/10.3390/s17122720
  6. G. Ivan, B. Marcos, and R. Ezequiel, "Improved Detection of Small Objects in Road Network Sequences using CNN and Super Resolution," Expert Systems, Vol. 39, No. 2, pp. 1-17, 2021. doi: https://doi.org/10.1111/exsy.12930
  7. R. Sheng, L. Jianqi, T. Tianyi, P. Yibo, and J. Jian, "Towards Efficient Video Detection Object Super-Resolution with Deep Fusion Network for Public Safety," Security and Communication Networks, Vol. 1, pp. 1-14, 2021. doi: https://doi.org/10.1155/2021/9999398
  8. W. Xinqing, H. Xia, X. Feng, L. Yuyang, H. Xiaodong, and S. Pengyu, "Multi-Object Detection in Traffic Scenes Based on Improved SSD," Electronics, Vol. 7, No. 11, pp. 1-28, 2018. doi: https://doi.org/10.3390/electronics7110302
  9. C. Luc, P. Tan, and L. Sebastien, "Small Object Detection in Remote Sensing Images Based on Super-Resolution with Auxiliary Generative Adversarial Networks," Remote Sensing, vol. 12, No. 19, pp. 1-19, Sep. 2020. doi: https://doi.org/10.3390/rs12193152
  10. W. Yunyan, W. Huaxuan, S. Luo, P, Chen, and Y. Zhiwei, "Detection of Plane in Remote Sensing Images using Super-resolution," Plosone, Vol. 17, No. 4, pp. 1-19, Apr. 2022. doi: https://doi.org/10.1371/journal.pone.0265503
  11. S. Ren, K. He, R. Girshick, and J. Sun, "Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks," IEEE Trans. Pattern Anal. Mach. Intell., Vol. 39, No. 6, pp. 1137-1149, June 2017. doi: https://doi.org/10.1109/TPAMI.2016.2577031
  12. M. Mostofa, S. Ferdous, B. Riggan, and N. Nasrabadi, "Joint-SRVDNet: Joint Super Resolution and Vehicle Detection Network," IEEE Access, Vol. 8, pp. 82306-82319, May, 2020. doi: https://doi.org/10.1109/ACCESS.2020.2990870
  13. L. Bee, S. Sanghyun, K. Heewon, N. Seungjun, L. Kyoungmu, "Enhanced Deep Residual Networks for Single Image Super-Resolution," In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Honolulu, HI, USA, pp. 21-26, June 2017. doi: https://doi.org/10.48550/arXiv.1707.02921
  14. C. Chen, J. Zhong, and Y. Tan, "Multiple-oriented and Small Object Detection with Convolutional Neural Networks for Aerial Image," Remote Sensing, Vol. 11, No. 18, pp. 1-23, 2019. doi: https://doi.org/10.3390/rs11182176
  15. M. Yogendra, M. Arvind, and K. Seol, "Single Image Super-Resolution Using Deep Residual Network with Spectral Normalization," 17th International Conference on Multimedia Information Technology and Application, Jeju, Korea, pp. 1-2, July 2021.