Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography (한국측량학회지)

Korean Society of Surveying, Geodesy, Photogrammetry and Cartography (한국측량학회)
권호 표지
DOI QR코드

DOI QR Code

Analysis of the Effect of Learned Image Scale and Season on Accuracy in Vehicle Detection by Mask R-CNN

Mask R-CNN에 의한 자동차 탐지에서 학습 영상 화면 축척과 촬영계절이 정확도에 미치는 영향 분석

  • Choi, Jooyoung (Dept. of Environmental Health Science, Dept. of Technology Fusion Engineering, Konkuk University) ;
  • Won, Taeyeon (Dept. of Advanced Technology Fusion, Konkuk University) ;
  • Eo, Yang Dam (Dept. of Civil and Environmental Engineering, Konkuk University)
  • Received : 2022.01.06
  • Accepted : 2022.02.17
  • Published : 2022.02.28
Download PDF
⟨ Previous Next ⟩

Abstract

In order to improve the accuracy of the deep learning object detection technique, the effect of magnification rate conditions and seasonal factors on detection accuracy in aerial photographs and drone images was analyzed through experiments. Among the deep learning object detection techniques, Mask R-CNN, which shows fast learning speed and high accuracy, was used to detect the vehicle to be detected in pixel units. Through Seoul's aerial photo service, learning images were captured at different screen magnifications, and the accuracy was analyzed by learning each. According to the experimental results, the higher the magnification level, the higher the mAP average to 60%, 67%, and 75%. When the magnification rates of train and test data of the data set were alternately arranged, low magnification data was arranged as train data, and high magnification data was arranged as test data, showing a difference of more than 20% compared to the opposite case. And in the case of drone images with a seasonal difference with a time difference of 4 months, the results of learning the image data at the same period showed high accuracy with an average of 93%, confirming that seasonal differences also affect learning.

본 연구에서는 딥러닝 객체탐지 기법의 정확도 향상을 위해 항공사진과 드론 영상을 대상으로 확대율 조건과 계절요인이 탐지정확도에 미치는 영향을 실험을 통해 분석하였다. 딥러닝 객체탐지기법 중 빠른 학습 속도와 높은 정확도를 나타내는 Mask R-CNN을 사용하여 탐지대상인 자동차를 픽셀 단위로 탐지하고자 하였다. '서울시 항공사진서비스'를 통해 화면 확대 레벨을 달리하며 학습 영상을 캡처하고 각각을 학습하여 정확도를 분석하였다. 실험결과에 따르면 확대 레벨이 높아질수록 mAP 평균이 60%, 67%, 75%로 높아졌다. 데이터 세트의 train, test 데이터의 확대율을 엇갈려서 배치한 경우에는 확대율이 매우 낮은 경우를 제외하고 저배율의 데이터를 train 데이터로, 고배율의 데이터를 test 데이터로 배치하였을 때 높은 mAP로 반대의 경우보다 20% 이상 차이를 보였다. 그리고 4개월의 시차로 계절적 차이를 두고 촬영한 드론 영상의 경우, 같은 시기 영상자료 학습결과가 평균 93%로 높은 정확도를 나타내어 계절적 차이도 학습에 영향을 주는 것을 확인되었다.

Keywords

References

  1. Girshick, R., Donahue, J., Darrell, T., and Malik, J. (2014), Rich feature hierarchies for accurate object detection and semantic segmentation, IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 24-27 June, Columbus, Ohio, pp. 580-587.
  2. Gu, H.I. (2018), Artificial intelligence and deep learning trends, The Korean Institute of Electrical Engineers, Vol. 67, No. 7, pp. 7-12. (in Korean)
  3. He, K., Gkioxari, G., Dollar, P., and Girshick, R. (2017), Mask R-CNN, IEEE International Conference on Computer Vision (ICCV), 22-29 October, Venice, Italy, pp. 2961-2969.
  4. Jeong, J.Y., Cho, W.S., Chang, H.J., and Jeong, J.W. (2008), A Study on Object-based Change Detection Using Aerial LiDAR Data, The Korean Society of Remote Sensing Spring Conference-2008, 21 March, Korea, pp. 95-100.
  5. Kim, J., Song, Y.H., and Lee, W.K. (2021), Accuracy analysis of Multi-series Phenological Landcover Classification Using U-Net-based Deep Learning Model -Focusing on the Seoul, Republic of Korea-, Korean Journal of Remote Sensing, Vol. 37, No. 3, pp. 409-418. (in Korean with English abstract) https://doi.org/10.7780/KJRS.2021.37.3.4
  6. Kim, Y.S., Kwak, G.H., Lee, K.D., Na, S.I., Park, C.W., and Park, N.W. (2018), Performance evaluation of machine learning and deep learning algorithms in crop classification: impact of hyper-parameters and training sample size, Korean Journal of Remote Sensing, Vol. 34, No. 5, pp. 811-827. (in Korean with English abstract) https://doi.org/10.7780/KJRS.2018.34.5.9
  7. Lecun, Y., Boser, B., Denker, J.S., Henderson, D., Howard, R.E., Hubbard, and W., Jackel, L.D., (1989), Backpropagation applied to handwritten zip code recognition, Neural Computation, Vol. 1, No. 4, pp. 541-551. https://doi.org/10.1162/neco.1989.1.4.541
  8. Lee, M.H., Nam, K.W., and Lee, C.W. (2019), Crack Detection on the Road in Aerial Image using Mask R-CNN, Journal of the Korea Industrial Information Systems Research, Vol. 24, No. 3, pp. 23-29. (in Korean with English abstract) https://doi.org/10.9723/JKSIIS.2019.24.3.023
  9. Lee, S.G., Park, Y.S., Lee, G.S., Lee, J.Y., and Lee, S.H. (2013), An automatic object extraction method using color features of object and background in image, The Journal of Digital Policy & Management, Vol. 11, No. 12, pp. 459-465. (in Korean with English abstract)
  10. Park, J.C., Son, S.B., Lee, S.H., Jung, J.U., Park, Y.J., and Oh, H.S. (2021), Deep Ensemble based Object Detection from Aerial Images, Journal of Institute of Control, Robotics and Systems, Vol. 27, No. 12, pp. 944-952. (in Korean with English abstract) https://doi.org/10.5302/J.ICROS.2021.21.0135
  11. Redmon, J., Divvala, S., Girshick, R., and Farhadi, A. (2016), You only look once: unified, real-time object detection, 29th IEEE Conference on Computer Vision and Pattern Recognition, 27-30 June, Las Vegas, Nevada, USA, pp. 779-788.
  12. Ren, S., He, K., Girshick, R., and Sun, J. (2015), Faster R-CNN: towards real-time object detection with region proposal networks, IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 39, No. 6, pp. 91-99.
  13. Russakovsky, O., Deng, J., Su, H., Krause, J., Satheesh, S., Ma, S., Huang, Z., Karpathy, A., Khosla, A., Bernstein, M., Berg, A.C., and Fei-Fei, L. (2015), Imagenet large scale visual recognition challenge, International Journal of Computer Vision, Vol. 115, No. 3, pp. 211-252. https://doi.org/10.1007/s11263-015-0816-y
  14. Simonyan, K. and Zisserman, A. (2015), Very deep convolutional networks for large-scale image recognition, Computer Vision and Pattern Recognition, Cornell University, Ithaca, NY, USA.
  15. Song, J.Y., Won, T.Y., Jo, S.M., Eo, Y.D., Park, S.Y., Shin, S.H., Park, J.S., and Kim, C.J. (2021), Comparative experiment of cloud classification and detection of aerial image by deep learning, Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography, Vol. 39, No. 6, pp. 351-360. (in Korean with English abstract) https://doi.org/10.7848/KSGPC.2021.39.6.351
  16. Szegedy, C., Liu, W., Jia, Y., Sermanet, P., Reed, S., Anguelov, D., Erhan, D., Vanhoucke, V., and Rabinovich, A. (2015), Going deeper with convolutions, 2015 IEEE Conference on Computer Vision and Pattern Recognition, 7-12 June, Boston, MA, USA, pp. 1-9.
  17. Youn, H.J., Lee, M.H., Jeong, Y.S., Lee, C.W., Lee, H.S., and Jo, J.W. (2019), A Study on Car Detection in Road Surface Using Mask R-CNN in Aerial Image, Korea Institute of information and Communication Engineering Spring Conference-2019, 23 May, Korea, pp. 71-73.