A HYBRID SYSTEM FOR VALIDATING CROWD-SOURCED REPORTS ABOUT RAMPS BASED ON CNN AND HITL

Zima Volodymyr; Ilya Kobylin; Oleksandra Putiatina

doi:10.36074/grail-of-science.15.05.2026.116

A HYBRID SYSTEM FOR VALIDATING CROWD-SOURCED REPORTS ABOUT RAMPS BASED ON CNN AND HITL

Authors

Zima Volodymyr Kharkiv National University of Radio Electronics, Ukraine https://orcid.org/0009-0005-5223-5228
Ilya Kobylin Kharkiv National University of Radio Electronics, Ukraine https://orcid.org/0000-0002-4552-9616
Oleksandra Putiatina Kharkiv National University of Radio Electronics, Ukraine https://orcid.org/0000-0003-4853-7125

DOI:

https://doi.org/10.36074/grail-of-science.15.05.2026.116

Keywords:

active learning, accessibility of the urban environment, crowdsourcing, low-mobility population groups, ramp, Human-in-the-Loop, YOLO, convolutional neural network

Summary

The article is devoted to the development and experimental verification of a hybrid pipeline for validating crowdsourced reports on accessibility elements of urban infrastructure, in particular ramps and lowered curbs. The proposed system combines automatic image classification by a convolutional neural network (CNN) with a Human-in-the-Loop (HITL) mechanism: reports with a confidence level below a threshold value are directed for verification to a specialist. A comparative analysis of YOLOv8n, YOLOv8m and ResNet-50 was conducted on the Sidewalk Accessibility v3.0 dataset (2603 images). YOLOv8m showed the best detection metrics: mAP@50 = 0.446, Precision = 0.538, Recall = 0.452. The HITL pipeline with a threshold of 0.50 automatically processes 66.4% of images and directs 33.6% to a human. Additional training on 73 verified samples improved mAP@50 to 0.478 (+7.2%), Precision to 0.596 (+10.8%). The concept of "accessibility trails" based on GPS tracks of low-mobility users is proposed.

Downloads

PDF

Downloads

Download data is not yet available.

License

References

Goodchild, M. F. (2007). Citizens as sensors: the world of volunteered geography. GeoJournal, 69(4), 211–221. https://doi.org/10.1007/s10708-007-9111-y DOI: https://doi.org/10.1007/s10708-007-9111-y

Hara, K., Sun, J., Moore, R., Jacobs, D., & Froehlich, J. (2014). Tohme. У UIST '14: The 27th Annual ACM Symposium on User Interface Software and Technology. ACM. https://doi.org/10.1145/2642918.2647403 DOI: https://doi.org/10.1145/2642918.2647403

Hosseini, M., Miranda, F., Lin, J., & Silva, C. T. (2022). CitySurfaces: City-scale semantic segmentation of sidewalk materials. Sustainable Cities and Society, 79, 103630. https://doi.org/10.1016/j.scs.2021.103630 DOI: https://doi.org/10.1016/j.scs.2021.103630

Kulkarni, M., Li, C., Ahn, J. J., Ma, K. O. Y., Zhang, Z., Saugstad, M., Wu, K., Eisenberg, Y., Novack, V., Chamberlain, B., & Froehlich, J. E. (2023). BusStopCV: A Real-time AI Assistant for Labeling Bus Stop Accessibility Features in Streetscape Imagery. У ASSETS '23: The 25th International ACM SIGACCESS Conference on Computers and Accessibility. ACM. https://doi.org/10.1145/3597638.3614481 DOI: https://doi.org/10.1145/3597638.3614481

Moreira, R. S., Moita, S., Torres, J. M., Gouveia, F., Dinis, M. A. P., Ferreira, D., Araújo, M., & Guerreiro, M. J. S. (2025). Automating City Accessibility Constraints Mapping Through AI‐Assisted Scanning of Street View Imagery. IET Smart Cities, 7(1). https://doi.org/10.1049/smc2.70020 DOI: https://doi.org/10.1049/smc2.70020

Mosqueira-Rey, E., Hernández-Pereira, E., Alonso-Ríos, D., Bobes-Bascarán, J., & Fernández-Leal, Á. (2022). Human-in-the-loop machine learning: a state of the art. Artificial Intelligence Review. https://doi.org/10.1007/s10462-022-10246-w DOI: https://doi.org/10.1007/s10462-022-10246-w

O'Meara, J. S., Hwang, J., Wang, Z., Saugstad, M., & Froehlich, J. E. (2025). RampNet: A Two-Stage Pipeline for Bootstrapping Curb Ramp Detection in Streetscape Images from Open Government Metadata. У 2025 IEEE/CVF International Conference on Computer Vision Workshops (ICCVW) (с. 6715–6724). IEEE. https://doi.org/10.1109/iccvw69036.2025.00693 DOI: https://doi.org/10.1109/ICCVW69036.2025.00693

Saha, M., Saugstad, M., Maddali, H. T., Zeng, A., Holland, R., Bower, S., Dash, A., Chen, S., Li, A., Hara, K., & Froehlich, J. (2019). Project Sidewalk. У CHI '19: CHI Conference on Human Factors in Computing Systems. ACM. https://doi.org/10.1145/3290605.3300292 DOI: https://doi.org/10.1145/3290605.3300292

Senaratne, H., Mobasheri, A., Ali, A. L., Capineri, C., & Haklay, M. (. (2016). A review of volunteered geographic information quality assessment methods. International Journal of Geographical Information Science, 31(1), 139–167. https://doi.org/10.1080/13658816.2016.1189556 DOI: https://doi.org/10.1080/13658816.2016.1189556

Wang, K., Zhang, D., Li, Y., Zhang, R., & Lin, L. (2017). Cost-Effective Active Learning for Deep Image Classification. IEEE Transactions on Circuits and Systems for Video Technology, 27(12), 2591–2600. https://doi.org/10.1109/tcsvt.2016.2589879 DOI: https://doi.org/10.1109/TCSVT.2016.2589879

Weld, G., Jang, E., Li, A., Zeng, A., Heimerl, K., & Froehlich, J. E. (2019). Deep Learning for Automatically Detecting Sidewalk Accessibility Problems Using Streetscape Imagery. У ASSETS '19: The 21st International ACM SIGACCESS Conference on Computers and Accessibility. ACM. https://doi.org/10.1145/3308561.3353798 DOI: https://doi.org/10.1145/3308561.3353798

Wu, X., Xiao, L., Sun, Y., Zhang, J., Ma, T., & He, L. (2022). A survey of human-in-the-loop for machine learning. Future Generation Computer Systems. https://doi.org/10.1016/j.future.2022.05.014. DOI: https://doi.org/10.1016/j.future.2022.05.014