An Explainable Smartphone-Based Deep Learning and Geospatial Framework for Road-Defect Surveillance and Maintenance Prioritisation in Rwanda: A Synthetic Proof of Concept
DOI:
https://doi.org/10.31305/ijmds.v15n08.002Keywords:
computer vision, deep learning, explainable artificial intelligence, road maintenance, smartphone sensing, RwandaAbstract
Road infrastructure is central to mobility, public safety, access to services and local economic development. Nevertheless, recurrent pavement defects, drainage obstruction and rain-related surface deterioration can outpace periodic manual inspection, particularly where inspection resources are constrained. This paper proposes a smartphone-based deep-learning and geospatial framework for detecting, classifying and prioritising road defects in Rwanda. The framework combines vehicle-mounted smartphone imagery, global positioning data, lightweight object detection and segmentation models, civil engineering severity rules and a geospatial maintenance dashboard. A reproducible synthetic proof of concept was implemented using 600 training images and 200 test images across normal-road, pothole, crack and standing-water classes. A two-hidden-layer neural net-work achieved 79.0% test accuracy and a 79.3% macro F1-score. Potholes obtained the strongest F1-score (88.4%), whereas crack recall was limited to 66.0%, exposing a mate-rial false-negative risk. These results establish software functionality but do not estimate performance on Rwandan roads. The paper synthesises computer-vision research with infrastructure-management and development perspectives and specifies a locally grounded data-collection, annotation, validation, explainability and governance protocol. Its principal contribution is not merely automated defect recognition, but a decision-support architecture linking low-cost sensing to accountable public maintenance. The framework and accompanying programme offer a feasible basis for field research, municipal piloting and scalable digital road stewardship in Rwanda and comparable low-resource settings.
References
[1] Africa Press. (2025, October 12). Sonatube, Gishushu, Kwa Lando junctions set for over-haul. https://www.africa-press.net/rwanda/all-news/sonatube-gishushu-kwa-lando-junctions-set-for-overhaul
[2] Africa Press. (2026, April 18). Rwanda acquires high-tech scanner for road quality monitoring. https://www.africa-press.net/rwanda/photo/rwanda-acquires-hig h-tech-scanner-for-road-quality-monitoring
[3] Arya, D., Maeda, H., Ghosh, S. K., Toshniwal, D., Mraz, A., Kashiyama, T., & Sekimoto, Y. (2021). Transfer learning-based road damage detection for multiple countries. Automation in Construction, 128, 103750. https://doi.org/10.1016/j.autcon.2021.1037 50 DOI: https://doi.org/10.1016/j.autcon.2021.103935
[4] Heitzmann, P. (2022). A computer vision-assisted approach to automated real-time road infrastructure management. arXiv. https://doi.org/10.48550/arXiv.2202.13285
[5] Li, Y., Yin, C., Lei, Y., Zhang, J., & Yan, Y. (2024). RDD-YOLO: Road damage detection algorithm based on improved You Only Look Once version 8. Applied Sciences, 14(8), 3360. https://doi.org/10.3390/app14083360 DOI: https://doi.org/10.3390/app14083360
[6] Lundberg, S. M., & Lee, S.-I. (2017). A unified approach to interpreting model predictions. Advances in Neural Information Processing Systems, 30, 4765-4774.
[7] Ministry of Infrastructure. (2022, October 27). Rwanda Urban Development Project car-ries on enhancing sustainable infrastructure and services in secondary cities. https://www.mininfra.gov.rw/updates/news-details/rwanda-urban-development-project-carries-on-enhancing-sustainable-infrastructure-and-services-in-secondary-cities
[8] Ministry of Infrastructure. (2024, March 14). Transforming lives and connecting com-munities: Celebrating the success of Rwanda’s Urban Development Project. https://www.mininfra.gov.rw/updates/news-details/transforming-lives-and-connecting-communities-celebrating-the-success-of-rwanda’s-urban-development-project
[9] The New Times. (2024, October 1). Careless driving responsible for 50% of Rwanda’s traffic accidents. https://www.newtimes.co.rw/article/20573/news/rwanda/c areless-driving-responsible-for-50-of-rwandas-traffic-accidents
[10] The New Times. (2026, June 28). Rwanda launches road model to boost resilience in disaster-prone areas. https://www.newtimes.co.rw/article/36936/news/featur ed/photos-rwanda-launches-road-model-to-boost-resilience-in-disaste r-prone-areas
[11] Niyishaka, P. (n.d.). Detecting cherry-picking and quantifying uncertainty in road-accident analytics for reliable public-safety decisions. University of Michigan MIDAS. https://midas.umich.edu/directory/niyishaka-patrick/
[12] Ramesh, A., Nikam, D., Balachandran, V. N., Guo, L., Wang, R., Hu, L., Comert, G., & Jia, Y. (2022). Cloud-based collaborative road damage monitoring with deep learning and smartphones. Sustainability, 14(14), 8682. https://doi.org/10.3390/su14148682 DOI: https://doi.org/10.3390/su14148682
[13] Rwanda Information Society Authority. (2021). Data protection and privacy law. https://www.risa.gov.rw/data-protection-and-privacy-law
[14] Rwanda National Police. (2025a). Gerayo Amahoro: Road accident victims narrate the ordeal of living with lifetime injuries, advise road users to exercise caution. https://police.gov.rw/media/news-detail/news/gerayo-amahoro-road-accident-victims-narrate-the-ordeal-of-living-with-lifetime-injuries-advise-ro/
[15] Rwanda National Police. (2025b, April 15). Police remind motorists to be cautious as heavy rains intensify. https://police.gov.rw/media/news-detail/news/polic e-remind-motorists-to-be-cautious-as-heavy-rains-intensify/
[16] Rwanda Transport Development Agency. (n.d.). National and Urban Roads Division. https://www.rtda.gov.rw/about/divisions-and-units/national-and-urban-roadsdivision
[17] Rwanda Transport Development Agency. (2025, October 30). RTDA Board of Directors conducts field visit to key transport infrastructure projects across Rwanda. https://ww w.rtda.gov.rw/updates/news-details/rtda-board-of-directors-conduct s-field-visit-to-key-transport-infrastructure-projects-across-rwanda
[18] Selvaraju, R. R., Cogswell, M., Das, A., Vedantam, R., Parikh, D., & Batra, D. (2020). Grad-CAM: Visual explanations from deep networks via gradient-based localization. International Journal of Computer Vision, 128, 336-359. https://doi.org/10.1007/s1 1263-019-01228-7 DOI: https://doi.org/10.1007/s11263-019-01228-7
[19] Y. Hu, N. Chen, Y. Hou, X. Lin, B. Jing, et al., “Lightweight deep learning for real-time road distress detection on mobile devices,” Nature Communications, vol. 16, Art. no. 4212, May 2025, doi: 10.1038/s41467-025-59516-5. DOI: https://doi.org/10.1038/s41467-025-59516-5
[20] Wang, C., & Yan, M. (2023). YOLO-LWNet: A lightweight road damage object detection network for mobile terminal devices. Sensors, 23(6), 3268. https://doi.org/10.339 0/s23063268 DOI: https://doi.org/10.3390/s23063268
[21] Wang, W., Yu, X., Jing, B., Tang, Z., Zhang, W., Wang, S., Xiao, Y., Li, S., & Yang, L. (2025). YOLO-RD: A road damage detection method for effective pavement mainte-nance. Sensors, 25(5), 1442. https://doi.org/10.3390/s25051442 DOI: https://doi.org/10.3390/s25051442
[22] Xinhua. (2025, February 12). 20 killed, dozens injured in road accident in northern Rwanda. http://english.news.cn/africa/20250212/f199c893680542079a acb2f36a15bce5/c.html
[23] Xu, L., Fu, K., Ma, T., Tang, F., & Fan, J. (2024). Automatic detection of urban pave-ment distress and dropped objects with a comprehensive dataset collected via smartphone. Buildings, 14(6), 1546. https://doi.org/10.3390/buildings14061546 DOI: https://doi.org/10.3390/buildings14061546
