Automated Traffic Counting system
Posted on Sunday, November 26th, 2023
Client | Stephen Gagne [MakerSpace] |
Professor(s) | Abdullah Kadri |
Program | Computer Engineering Technology [CET] |
Students | Mark Sardar, Erik Kavanagh, Hiran Samarasinghe, Aeham Alobidat, Jordon Kent |
Project Description:
Our project aimed to address the need for a robust 24/7 surveillance system employing two web cameras for real-time person detection and tracking. The client required a solution that could accurately monitor and log hourly and total detections, offering a comprehensive view of the surveillance environment.
Our team implemented HAAR cascade classifiers for upper body detection and KCF trackers for person tracking in the captured frames. The system not only logged detection counts for each camera but also maintained a timestamped Excel file, providing hourly and cumulative counts for efficient monitoring and analysis. The user-friendly interface displayed live feeds from both cameras, with detection counts updated every 30 minutes.
Throughout the development process, we encountered challenges related to detection accuracy, leading us to explore potential enhancements. We learned to balance the intricacies of computer vision techniques and real-world application, gaining insights into the dynamic nature of surveillance scenarios.
Incorporating user feedback and iterative testing, we refined the system to achieve optimal performance. While the current implementation serves as a foundational solution, we recognize the potential for further advancements. Future recommendations include exploring deep learning models for enhanced accuracy, integrating real-time alerts, and optimizing the system for diverse environmental conditions.
Our project represents a significant step toward an intelligent surveillance system, laying the groundwork for real-time monitoring applications. The combination of HAAR cascade classifiers and KCF trackers offers a versatile solution applicable in various scenarios. The project not only met the client’s requirements for continuous monitoring but also provided valuable insights for future enhancements.
Explore our interactive demo to experience the seamless integration of detection and tracking in a live surveillance environment. We invite you to witness the potential of our system in revolutionizing real-time monitoring, paving the way for advancements in surveillance technology.