An android application to detect and classify pothole based on image, and then report to the database
Presentation.of.Application.mp4
Pothole Detection and Severity Classification: A Mobile Application of Image Recognition and Machine Learning for Improved Road Maintenance
This project presents a novel approach to address urban road maintenance issues, particularly focusing on pothole detection and classification in the United Kingdom. Utilising advancements in machine learning, specifically Convolutional Neural Networks (CNN), combined with smartphone technologies, this system aims to revolutionise the traditional methods of pothole reporting, making them more efficient and user-friendly.
Machine Learning Model: Incorporates SSD with MobileNetV2 for efficient object detection and classification of potholes.
Mobile Application: User-friendly interface designed using Flutter SDK, allowing real-time image capture, upload, pothole detection, and classification.
Data Handling: Uses Firebase Realtime Database for secure data storage, ensuring GDPR compliance and data privacy.
Ensure Flutter SDK and dependencies (TensorFlow, NumPy, OpenCV, and Pandas) are installed. Clone the repository and build the application. Run the app on a compatible device to capture or upload images for pothole detection.
This project is licensed under the Creative Commons Attribution-NonCommercial 4.0 International License. For more details, see the LICENSE file.
This project is not just a demonstration of technical skills in machine learning and mobile app development, but also an exploration into academic writing and publication. Although my initial submission to arXiv was not accepted, this experience has been invaluable in understanding the academic publication process and has provided insights for future research dissemination.
Feel free to connect with me on LinkedIn (linkedin.com/in/pserdarakin/) to discuss this project, potential collaborations, or job opportunities. My journey from conceptualization to development and attempted publication of this work illustrates my commitment to continual learning and professional growth.
I want to give my deepest thanks to my supervisor, Prof Xiaohong Gao. She has been a constant source of support and has guided me throughout this entire dissertation. Her wisdom and experience have been very important in helping me do my best work.
I am also very thankful to my family, who have always been there for me during my studies and while working on this thesis. Their ongoing support and belief in me have meant a lot, and I could not have done it without them.
29 September 2023 Serdar Akin