DOI QR코드

DOI QR Code

Analysis of Deep Learning-Based Pedestrian Environment Assessment Factors Using Urban Street View Images

도시 스트리트뷰 영상을 이용한 딥러닝 기반 보행환경 평가 요소 분석

  • 황지연 (군산대학교 컴퓨터정보공학과) ;
  • 최철웅 (군산대학교 컴퓨터정보공학과) ;
  • 남광우 (군산대학교 컴퓨터정보공학과) ;
  • 이창우 (군산대학교 컴퓨터정보공학과)
  • Received : 2023.09.26
  • Accepted : 2023.11.19
  • Published : 2023.12.31

Abstract

Recently, as the importance of walking in daily life has been emphasized, projects to guarantee walking rights and create a pedestrian environment are being promoted throughout the region. In previous studies, a pedestrian environment assessment was conducted using Jeonju-si road images, and an image comparison pair data set was constructed. However, data sets expressed in numbers have difficulty in generalizing the judgment criteria of pedestrian environment assessors or visually identifying the pedestrian environment preferred by pedestrians. Therefore, this study proposes a method to interpret the results of the pedestrian environment assessment through data visualization by building a web application. According to the semantic segmentation result of analyzing the walking environment components that affect pedestrian environment assessors, it was confirmed that pedestrians did not prefer environments with a lot of "earth" and "grass," and preferred environments with "signboards" and "sidewalks." The proposed study is expected to identify and analyze the results randomly selected by participants in the future pedestrian environment evaluation, and believed that more improved accuracy can be obtained by pre-processing the data purification process.

최근 일상생활 속 보행의 중요성이 강조되면서 보행권 보장 및 보행환경 조성을 위한 사업이 지역 곳곳에서 추진되고 있다. 선행 연구에서는 전주시 도로 이미지를 사용하여 보행환경 평가를 진행하고, 이미지 비교 쌍 데이터 세트를 구축하였다. 하지만 숫자로 표현된 데이터 세트는 보행환경 평가자들의 판단 기준을 일반화하거나 보행자가 선호하는 보행환경을 시각적으로 파악하기에 어려움이 존재한다. 따라서 본 연구는 웹 애플리케이션을 구축하여 데이터 시각화를 통해 보행환경 평가의 결과를 해석하는 방법을 제안한다. 의미론적 분할 결과를 활용하여 보행환경 평가자에게 영향을 미치는 보행환경 구성 요소를 분석한 결과, 보행자는 주로 'earth'와 'grass'가 많은 환경을 선호하지 않았고, 'signboard'와 'sidewalk'를 가진 환경을 선호하는 것으로 확인하였다. 제안된 연구는 향후 보행환경 평가의 참여자가 임의로 선택한 결과를 파악하고 분석할 수 있을 것으로 기대하며, 데이터에 대한 정제과정을 전처리로 수행함으로써 좀 더 향상된 정확도를 얻을 수 있을 것으로 판단한다.

Keywords

Acknowledgement

이 논문은 국토교통부/국토교통과학기술진흥원의 지원에 의해 연구되었음(과제번호 RS-2022-00143336).

References

  1. AURUM, Safe Walking Environment Improvement Project, https://www.aurum.re.kr/Policy/PolicyOffice.aspx?pcode=A01(Accessed on Nob, 21th, 2023)
  2. Dubey, A., Naik, N., Parikh, D., Raskar, R., & Hidalgo, C. A. (2016). Deep learning the city: Quantifying urban perception at a global scale. In Computer Vision-ECCV 2016: 14th European Conference, Amsterdam, The Netherlands, October 11-14, 2016, Proceedings, Part I 14 (pp. 196-212). Springer International Publishing. https://doi.org/10.1007/978-3-319-46448-0_12.
  3. Hyunglak Kim. (2016). Improvement of Crowdsourcing based Software Development Process, Korea Computer Congress 2016, 2016(6), 654-656.
  4. Jiyoung Park, Youngok Kang, and Jiyeon Kim. (2022). Development of Walkability Evaluation Index Using StreetView Image and Semantic Segmentation, Journal of the Korean Cartographic Association, 22(1), 53-68. https://doi.org/10.16879/jkca.2022.22.1.053.
  5. KiChun Lee, KwangWoo Nam, and ChangWoo Lee. (2022). A Study on the Walkability Scores in Jeonju City Using Multiple Regression Models, Journal of the Korea Industrial Information Systems Research, 27(4), 1-10. https://doi.org/10.9723/jksiis.2022.27.4.001.
  6. Kihyun Yoo, Donggi Lee, Chang Woo Lee, and Kwang Woo Nam. (2022). Generating Pairwise Comparison Set for Crowed Sourcing based Deep Learning, Journal of the Korea Industrial Information Systems Research, 27(5), 1-11. http://dx.doi.org/10.9723/jksiis.2022.27.5.001.
  7. Kim, Ji Yeon and Kang, Young Ok. (2022). Development of a Deep Learning Model to Predict the Qualitative Evaluation of a Walking Environment based on Street View Images, Journal of Korean Society for Geospatial Information Science, 30(2), 45-56. http://dx.doi.org/10.7319/kogsis.2022.30.2.045.
  8. Lee Dong Hyun and Lee Chun Jae. (2012). A Study on the Applicability of Crowdsourcing for Cadastral Reform, Journal of the Korean Society of Cadastre, 28(2), 55-70. https://doi.org/10.22988/KSC.2012.28.2.005
  9. Lee, Kyung-Hwan and Ahn, Kun-Hyuck. (2008). An Empirical Analysis of Neighborhood Environment Affecting Resident's Walking-A Case study of 12 Areas in Seoul-, Journal of the Architectural Institute of Korea Planning & Design, 24(6), 293-302.
  10. Lee, Seung Young. (2020). Strategies for the Vitalization of Crowd Sourcing According to the Classification of Its Types, Journal of the Korean Society of Design Culture, 26(3), http://dx.doi.org/10.18208/ksdc.2020.26.3.307.
  11. Lieu, Seungjae et al., (2021). Analysis of Street Environmental Factors Affecting Subjective Perceptions of StreetScape Image in Seoul, Korea: Application of Deep Learning Semantic Segmentation and YOLOv3 Obeject Detection, Journal of Korea Planning Association, 56(2), 79-93. https://doi.org/10.17208/jkpa.2021.04.56.2.79.
  12. Seresinhe, Chanuki Illushka, Tobias Preis, and Helen Susannah Moat. (2017). Using deep learning to quantify the beauty of outdoor places, Royal Society open science, 4(7). https://doi.org/10.1098/rsos.170170.
  13. Suwon-cityhall. (2023), Suwon City holds campaign for 'BOOJ IRUN', https://www.suwon.go.kr/web/board/BD_board.view.do?bbsCd=1043&seq=20230904094532891 (Accessed on Sep, 07th, 2023)
  14. Wanjucoop. (2023) Recruitment of participants in the 2023 Jeonbuk Innovation City's Picking Practice Project, http://www.wanjucoop.com/bbs/notice/15781 (Accessed on Sep, 13th, 2023