RoadWatchAI: Intelligent Pothole Detection with Android Camera

References

[1] S. R. S. T. et al., "Advancements in real-time road damage detection: a comprehensive survey of methodologies and datasets," Journal of Real- Time Image Processing, vol. 22, no. 3, pp. 145- 162, 2025.

[2] S. S. et al., "Deep CNN and twin support vector machine based model for detecting potholes in road network," International Journal on Smart Sensing and Intelligent Systems, vol. 18, no. 2, pp. 78-95, 2025.

[3] A. K. et al., "IoT-Based System for Real-Time Recognition of Potholes and Humps with GPS Integration," presented at the IEEE International Conference on Smart Technologies, 2025, pp. 234-239.

[4] A. K. et al., "Advancements in pothole detection techniques: a comprehensive review and comparative analysis," Discover Artificial Intelligence, vol. 5, no. 12, pp. 1-28, 2025.

[5] A. Zhumadillayeva et al., "An intelligent hybrid YOLO–CNN–LSTM framework for real-time road infrastructure monitoring and analysis," Measurement, vol. 198, no. 111234, pp. 1-15, 2026.

[6] P. Vijay Ganesh, V. Sam Joshua, L. Saro Murugan, and S. Mahalakshmi, "Yolov9 Driven Internet Of Things Enabled Sustainable Solution For Intelligent Traffic Light Management System For Emergency Vehicles In Large Scale Urban Traffic," presented at the 2nd International Conference on Machine Learning and Autonomous Systems (ICMLAS 2025), 2025.

[7] Shebin Silvister, Dheeraj Komandur, Shubham Kokate, Aditya Khochare, Uday More, Vinayak Musale, and Avadhoot Joshi, "Deep Learning Approach to Detect Potholes in Real-Time using Smartphone," presented at the 2019 IEEE Pune Section International Conference (PuneCon), 2019, pp. 1-4.

[8] Zahid Hasan, Samsoon Nahar Shampa, Tasmia Rahman Shahidi, and Shahnewaz Siddique, "Pothole and Speed Breaker Detection Using Smartphone Cameras and Convolutional Neural Networks," presented at the 2020 IEEE Region 10 Symposium (TENSYMP), 2020, pp. 279- 282.

[9] Sachin B. Bhosale and Sivaram Ponnusamy, "Smart Pothole Detection and Reporting System: Real-Time Localization and Automated Alerts for Efficient Road Maintenance," EPJ Web of Conferences, vol. 343, no. 05012, pp. 1- 8, 2025.

[10] Real Time Pothole Detection System Using Deep Learning Techniques: A Systematic Review, presented at the 2024 4th International Conference on Ubiquitous Computing and Intelligent Information Systems (ICUIS), 2024, pp. 453-459.

[11] Pothole Severity Classification Using Structural Features Extracted from Segmented Images, presented at the 2024 Arab ICT Conference (AICTC), 2024, pp. 52-58.

[12] K. Zhang et al., "Road damage detection using deep learning with aerial imagery," IEEE Transactions on Intelligent Transportation Systems, vol. 25, no. 4, pp. 3210-3223, 2024.

[13] M. Rahman, L. Chen, and S. Patel, "Multi-scale convolutional neural networks for pothole detection in varying lighting conditions," Pattern Recognition Letters, vol. 168, pp. 45-53, 2023.

[14] R. Kumar and P. Singh, "Real-time road anomaly detection using smartphone sensors and edge computing," IEEE Sensors Journal, vol. 24, no. 8, pp. 12890-12901, 2024.

[15] T. Nakamura, H. Suzuki, and Y. Tanaka, "Automated pavement distress classification using transfer learning and ensemble methods," Journal of Computing in Civil Engineering, vol. 38, no. 2, pp. 04024005-04024018, 2024.

@article{1717869,
      author = {Rajesh Kanna S, Yogesh P},
      title = {RoadWatchAI: Intelligent Pothole Detection with Android Camera},
      journal = {Iconic Research And Engineering Journals},
      year = {2026},
      volume = {9},
      number = {11},
      pages = {2205-2213},
      issn = {2456-8880},
      url = {https://www.irejournals.com/formatedpaper/1717869.pdf},
      abstract = {This paper presents a novel multimodal deep learning framework for pothole detection that addresses the critical limitations of traditional approaches, which typically rely on single-modal data sources and suffer from high false positive rates and poor performance across varying road conditions. The proposed system integrates visual data from cameras, vibration data from accelerometers, and spatial information from GPS sensors through feature-level fusion using Multiple Scale Convolutional Neural Networks (MSCNN), enabling robust detection of both small and large potholes across diverse environmental conditions. To further enhance accuracy and interpretability, the framework employs an ensemble learning strategy incorporating a deep neural decision forest that extracts abstract features while maintaining interpretable decision boundaries, with hyperparameters optimized through a hybrid Genetic Algorithm and Gradient-Based Refinement approach. Experimental results demonstrate that this multimodal fusion approach significantly improves detection reliability compared to unimodal systems, effectively capturing pothole variations in size, road surface characteristics, and sensor noise conditions. The system's ability to cohesively integrate multiple sensor modalities through MSCNN feature extraction, combined with ensemble learning and advanced optimization techniques, provides a comprehensive solution for real-time road condition monitoring that can positively impact road maintenance operations and traffic safety management.},
      keywords = {Multimodal Deep Learning, Feature-Level Fusion, Multiple Scale Convolutional Neural Networks, Ensemble Learning, Real-Time Pothole Detection, Road Condition Monitoring},
      month = {May},
      doi = {https://doi.org/10.64388/IREV9I11-1717869}
  }