SafeWay – Smart Traffic Management System

International Journal of Computer Application ISSN 2250-1797 Available online on https://rspublication.com/ijca/ijca_index.htm Volume 15 Number. 6, 2025 DOI: 10.5281/zenodo.17574648 Original Article ©2025 RS Publication, [email protected] 36 SafeWay – Smart Traffic Management System Dr Subasree S 1 ,Adharsha S T 2 , Aro Marya Adam R 3 , Kanisgha Varssini C 4 , Meenashree Jagadeeshkumar 5 1 Dean Academics Head of department, Computer Science and Engineering (Cyber Security), Sri Shakthi Institute of Engineering and Technology, Coimbatore, Tamil Nadu, India 2-5 Student, Computer Science and Engineering (Cyber Security), Sri Shakthi Institute of Engineering and Technology, Coimbatore, Tamil Nadu, India International Journal of Computer Application https://rspublication.com/ijca/ijca_index.htm ISSN 2250-1797 ARTICLE INFO ABSTRACT ©2025 RS Publication Paper ID: IJCA690DC1D909CC2 Received: 2025-10-09 Published: 2025-11-10 DOI: https://dx.doi.org/ 10.5281/zenodo.1757 4648 Page No: 36-43 SafeWay is a cutting-edge approach to traffic management that aims to reduce urban traffic congestion, enhance roadway safety, and prioritize emergency vehicles and pedestrians based on artificial intelligence (AI) and machine learning (ML). SafeWay uses an object detector inspired by YOLO to detect vehicles and pedestrians in real-time while mixing in a siren detector for emergency vehicles based on audio processing. SafeWay effectively modifies the timing of traffic signals, providing real-time notification to involved parties, including pedestrians, and ultimately increases vehicle/pedestrian coordination and decreases wait times to traverse the intersection. SafeWay will help urban settings with components related to pedestrian detection, crosswalk enforcement, and safe routes for a safer and more efficient movement of pedestrians and motor vehicles in an urban context while modernizing traffic management with AI. Keywords: Smart Traffic, Artificial Intelligence, Machine Learning, YOLO, Siren Detection, Emergency Vehicle, Traffic Management, SafeWay. Cite This Paper: Dr Subasree S, ,Adharsha S T, Aro Marya Adam R, Kanisgha Varssini C and Meenashree Jagadeesh kumar (2025). "SafeWay – Smart Traffic Management System". INTERNATIONAL JOURNAL OF COMPUTER APPLICATION (IJCA), vol. 15, no. 6, 2025, pp. 36-43. DOI: h4ps://dx.doi.org/10.5281/zenodo.17574648 International Journal of Computer Application ISSN 2250-1797 Available online on https://rspublication.com/ijca/ijca_index.htm Volume 15 Number. 6, 2025 DOI: 10.5281/zenodo.17574648 Original Article ©2025 RS Publication, [email protected] 37 2. Introduction Cities worldwide continue to face a multitude of significant difficulties related to the problems connected with traffic congestion, response times for emergency responders, and unsafe roadway designs for pedestrians. Every driver has encountered traffic signal systems that function with a timers system that does not account for the context of the condition on the road. As a resultant, drivers (or worse, waiting) often find themselves driving longer than they should for that traffic signal light are green, aggravating challenges for traffic flow. All the aforementioned characteristics feature the need for an intelligent traffic control system that not only monitors and acts on the conditions of the roadway but also considers a priority of emergency services and the expeditious and safe access for pedestrians. SafeWay addresses this problem leveraging Artificial Intelligence (AI) and Machine Learning (ML) and delivers a responsive and automatic traffic management system. The SafeWay system utilizes real-time audio based siren identification detection or recognition to determine the presence of fire trucks or ambulances and automatically adjusts the traffic signal accordingly allowing for vehicles to pass normally without stopping. SafeWay utilizes vision systems to identify pedestrians who may be in the crosswalk reducing unintentional impacts with vehicles. SafeWay increases urban roadway safety and mobility, enhances traffic efficiency, and provides green mobility initiatives by utilizing AI-based automation and multimodal detection of vehicles and pedestrians. 3. Literature Review Traffic management systems are increasingly utilizing Artificial Intelligence (AI) and Machine Learning (ML) techniques to improve mobility and safety in urban locations [I], [VII]. Traditional systems, such as adaptive signal control and vehicle detection, rely heavily on visual sensors that have limited performance during inclement weather or low visibility. To address these limitations, multiple studies have proposed hybrid AI-based approaches that utilize predictive algorithms and pattern recognition designed to support better decisions around traffic flow and response time to congestion [IV], [VII]. For example, Markov models and fuzzy clustering methods were used for short-term prediction of traffic patterns, and neural networks trained on extensive datasets showed improved adaptability of decisions for managing congestion [IV], [VII]. Additionally, multi-modal systems integrating visual and audio detection sought to enhance the identification of emergency vehicles in noisy and complex urban environments [V]. Although technology has come a long way, many of the advanced features are still dependent on independent modules and have limited functions for real-time interaction across the sensing, analysis, and decision-making aspects [VI], [X]. The SafeWay framework is designed to address these issues through the use of AI-based video and audio as a mechanism to detect and prioritize, provide real-time priority signal control for emergency vehicles to manage safety while ensuring traffic flow is optimized [I], [V], [VIII], [IX]. 4. Related Work The progress made in the field of Artificial Intelligence (AI) and sensing capabilities has led toward the establishment of intelligent transportation systems that are suitable replacements for traditional fixed-signal forms of control[I], [VII]. Although earlier adaptive approaches such as SCOOT and SCATS incorporated the use of sensor feedback to International Journal of Computer Application ISSN 2250-1797 Available online on https://rspublication.com/ijca/ijca_index.htm Volume 15 Number. 6, 2025 DOI: 10.5281/zenodo.17574648 Original Article ©2025 RS Publication, [email protected] 38 improve signal management, they were less capable of accommodating emergency vehicles during heavy congestion. Computer vision–based systems using computer image processing algorithms have been employed for vehicle and pedestrian detection, achieving high accuracy in optimal visibility conditions, but performing poorly in scenarios with adverse conditions such as rain, fog, or heavy traffic [III], [IV], [VII]. Audio-based systems, using Machine Learning (ML) systems with MFCCs and either CNN or SVM classifiers, have been employed for emergency siren detection, although these methods can be influenced by background traffic noise [V], [VIII]. In response to the above challenges, applications that use a multimodal detection approach combining audio and video data have been proposed to enhance detection reliability while reducing false alarms. SafeWay builds on these ideas and is an AI, multimodal, multi-sensory framework that utilizes real-time audio and visual analytics for improving emergency response, decreasing congestion, and increasing overall traffic safety in urban environments[I], [V], [VII], [VIII]. 5. Proposed Methodology The proposed SafeWay Smart Traffic Management System intends to update outdated traffic control systems by utilizing computer vision, deep learning, and real-time sensor fusion techniques. The system is designed to provide adaptive, intelligent and context-aware vehicle signal management based on real-world conditions like vehicle congestion, pedestrian presence, and emergency vehicle activity. In contrast to fixed timing signal control systems, the SafeWay system works by employing real-time vehicle camera and audio data to detect vehicles, pedestrians and emergency vehicle sirens. With this information, SafeWay more effectively controls signal timing for maintaining vehicle flow, safe pedestrian crossings, and ensures priority passage for emergency services. During traffic conditions with high density or limited visibility, the SafeWay system also engages audio-visual fusion to improve decision making with respect for accuracy and reliability.. Step 1: Gathering Data - In order to collect real-time data regarding the traffic volume and pedestrians movement along with sirens of emergency vehicles; this will be accomplished with sensors (microphones, cameras). Step 2: Data Preprocessing - Data preprocessing will consist of speaker noise removal, deletion of irrelevant items, and alignment of audio and video timestamps for clean and relevant data input. The intent is to obtain a clean and structured data set for more accurate inference into the models. Step 3: Detection Modules - SafeWay has three specialized detection models, automated via deep learning, that run concurrently: Emergency Vehicle Detection (Audio-based): Siren audio is used to generate Mel-spectrogram features that are input into a TensorFlow-lite neural network classifier that operates and executes in near real-time. X={x1,x2,...,xn}, Labels: Y={y1,y2,...,yn}, yi^=f(xi;θ) Vehicle Detection (Vision-based): The visual module employs a deep convolutional neural network (CNN) that has been trained on car, bus, and two-wheel datasets to detect and track vehicles. The model generates frame-wise bounding boxes and probability scores to estimate lane density for tracked vehicles. International Journal of Computer Application ISSN 2250-1797 Available online on https://rspublication.com/ijca/ijca_index.htm Volume 15 Number. 6, 2025 DOI: 10.5281/zenodo.17574648 Original Article ©2025 RS Publication, [email protected] 39 Pedestrian Detection (Vision-based): A lightweight deep learning model YOLO-11 identifies humans and performs detection on their movement in real-time with great accuracy. The model uses the detection to define pedestrian waiting areas, to assess pedestrian crossing movement, and track crowd sizes so SafeWay can intelligently extend green signals for pedestrians as appropriate. Step 4: Sensor Fusion - Combine the audio with visual detection to improve detection rates in crowded and noise ambient situations. Step 5: Dynamic Traffic Control - Dynamic signal control will utilize AI based predictive problem solving models to make real-time adjustments to traffic light signal patterns to create an open roadway for the emergency vehicle in both timely and cost-efficient way that would not unduly interfere with the flow of traffic patterns and the movement of regular traffic. Step 6: Continuous Monitoring and Prediction - SafeWay is always keeping an eye on the traffic and the environment for their real-time adaptability to be preserved. By using predictive models, it is possible to determine the next congestion patterns which then signal timings can be changed accordingly. Furthermore, the data collected from every cycle is not just stored but also analyzed with the aim of making the predictions more accurate in the future and improving the system's overall intelligence Fig.2: SafeWay System proposed methodology International Journal of Computer Application ISSN 2250-1797 Available online on https://rspublication.com/ijca/ijca_index.htm Volume 15 Number. 6, 2025 DOI: 10.5281/zenodo.17574648 Original Article ©2025 RS Publication, [email protected] 40 SafeWay will be designed through six phases. In regards to Phase 1, the real-time data will be recorded or tracked by sensing device telemetry (cameras, microphones, or vehicle sensors), using data on road traffic flow, weather, and sirens. After data is accumulated, it will then be subjected to Phase 2 preprocessing by denoising and deleting duplicates and inconsistencies to have an established clean and usable data set that would be relevant to input patterns. The third stage consists of detecting sirens using the audio modality. The audio input is processed into Melspectrogram features, which are then passed into a neural network (TensorFlow Lite) model for siren classification. On a mathematical level it could be expressed: Features: X = {x1, x2, ..., xn} Class labels: Y = {y1, y2, ..., yn} Predicted model: ŷi = f(xi; θ) These features are the incoming audio stream and the model is the neural network classification model. Here θ is the hyperparameter selected in the training stage. The fourth stage is Sensor Fusion: S = αA + (1 − α)V where the A is the audio confidence score discussed above, V is a visual confidence score discussed above and α is a weight of contribution of audio versus visual. The fifth stage is Dynamic Signal Control: D(t) = 1 (emergency vehicle detected) or 0 (normal) because this phase will have the dynamic traffic signal control for emergency vehicles. This is a continuous process as social media will continue to be updated in real-time as part of its functionality, to inform the emergency response and traffic situation models, and better respond to the occurrence of emergency vehicles. 6.Result and Discussions The Smart Traffic Management System uses cameras to monitor everything: cars in each lane, road congestion, and pedestrian movements. This data is gathered in real-time, after which smart algorithms make updates to the traffic lights in real-time that adjusts the signals to whatever is happening at the intersection. All of this creates better traffic flow and maintains traffic speed. Based on simulations and testing, it was demonstrated that the amount of traffic congestion and waiting time in a vehicle were both reduced in comparison to standard fixed-timed signal systems. The adaptive control logic adjusted signal timing dynamically so signals changed in real-time to fluctuation of traffic flow to prevent delays in low density lanes. The emergency vehicle detection module was found to reliably detect traditional sounds or visually detect emergency vehicle (ambulance, fire truck, etc.) to allow for a green signal that permitted rapid-moving emergency vehicles International Journal of Computer Application ISSN 2250-1797 Available online on https://rspublication.com/ijca/ijca_index.htm Volume 15 Number. 6, 2025 DOI: 10.5281/zenodo.17574648 Original Article ©2025 RS Publication, [email protected] 41 through the intersection. The system managed to maintain optimal traffic flow in all conditions, even during peak hour traffic. From a quantitative perspective, the average wait time for vehicles at busy intersections decreased significantly together with a clear reduction in the incidence of traffic jams. The study indicated that intelligent control algorithms augmented with real-time sensing represented a useful technique for optimizing urban traffic management systems. Fig.3 Average Emergency Vehicle Clearance Time per Intersection Fig 3 shows Average Emergency Vehicle Clearance Time per Intersection for two conditions. Without a smart system, the average clearance for an emergency vehicle at an intersection is 90 seconds. With the introduction of a "Smart System," the clearance time is significantly reduced to 30 seconds. That's a significant reduction - you have improved the clearance time by 60 seconds. The response time suggests that smart technology can significantly improve the speed and efficiency of emergency response. Fig.4: Average Pedestrian Wait Time per Intersection International Journal of Computer Application ISSN 2250-1797 Available online on https://rspublication.com/ijca/ijca_index.htm Volume 15 Number. 6, 2025 DOI: 10.5281/zenodo.17574648 Original Article ©2025 RS Publication, [email protected] 42 Fig 4 illustrates Average Pedestrian Wait Time over intersections. When not in use of a smart system, the pedestrian experienced an average wait time of 60 seconds to cross the street. However, this average wait time decreased to 35 seconds with the smart system "Take Away." The graphic clearly indicates reduced average wait time of approximately 25 seconds. The data would suggest improvements to pedestrian experience with smart intersection technology. The results of the study demonstrate strong support for the functionality of the proposed Smart Traffic Management System in ameliorating the shortcomings of traditional traffic signal operation. Fixed-time systems cannot adapt to changes in updated traffic volume, leading to congestion, fuel waste, and extended travel time. The proposed smart system, adjusts traffic signal timings dynamically, in minute intervals, based on live data. Using real-time sensing technology is key to tracking road conditions and managing traffic at intersections. Through the usage of data from sensors and cameras, the system increases situational awareness and allows for intelligent decision-making that is backed by data. Devices communicate with a central control unit, which coordinates the devices across multiple intersections and, in turn, reduces bottlenecks and improves traffic flow through nested road networks.The social impact of the system also initializes by detecting an emergency vehicle, where ambulances and fire/police services no longer have to manage their own flow through traffic signals like a standard vehicle. The function to enhance the flow for emergency vehicles minimizes the response time of service delivery, but also reduces accident risk for simply attracting manual intervention at signals. Further, the results demonstrate the ability to use the system as a smart city initiative. If using cloud storage, predictive analytics, and machine learning within a variation of the system. 7. Conclusion and Future work The SafeWay system is a smart and effective solution to the problems that urban traffic management is dealing with today. The combination of AI-based models and the audio and visual detection will allow the system to track the movements of cars, people, and emergency vehicles in real-time thus creating a very responsive and adaptive environment. The sensor fusion technique uses confidence scores from both modalities to identify the correct detection of emergency vehicles, therefore, minimizing delays and false alarms, for instance, ambulances, fire trucks, and police cars. The main advantage of the SafeWay system is dynamic traffic signal control, which permits the real-time modification of the traffic signal lights enabling the emergency vehicles to pass without any hindrance while the normal traffic is not affected. The system is learning and growing its reliability and adaptability to different traffic conditions continuously in real-time. It generates predictive recommendations to prevent congestion through intelligent AI models which are trained with optimized parameters for better traffic efficiency. To sum up, SafeWay signifies that AI-driven systems have the potential to not only enhance the safety of highways but also make them smarter by faster response times, better commuter experience, and urban mobility. International Journal of Computer Application ISSN 2250-1797 Available online on https://rspublication.com/ijca/ijca_index.htm Volume 15 Number. 6, 2025 DOI: 10.5281/zenodo.17574648 Original Article ©2025 RS Publication, [email protected] 43 SafeWay can be further extended in the future with more sophisticated detection for pedestrians, cyclists as well as public transport vehicles in the context of universal safety and inclusion. A multi-intersection coordination, predictive analytics for collision risk minimization, and adaptive route planning could be covered as well by the system so as to improve urban traffic networks on a large scale. These breakthroughs will propel SafeWay one step closer in being part of a fully automated and intelligent traffic management ecosystem.. 8. References I. C. Patange, K. G. Deekshitha, and N. S., Intelligent Real-Time Traffic Management System, Project Ref. No. 47S_BE_5487, Dept. of Electronics and Communication Engineering, PES University, Bangalore, India, External Title Approved (ETA) – 47th Series Student Project Programme (SPP), 2023–24, supervised by Prof. Manikandan J. II. Roboflow, Train Computer Vision Models, 2025. [Online]. Available: https://roboflow.com/ III. P. Kunekar, Y. Narule, R. Mahajan, S. Mandlapure, E. Mehendale, and Y. Meshram, Traffic Management System Using YOLO Algorithm, in Proc. Int. Conf. Recent Advances in Science and Engineering (RAiSE), Dubai, UAE, 4–5 Oct. 2023, Eng. Proc., Vol. 59(1), p. 210, Jan. 2024, doi: 10.3390/engproc2023059210. IV. A. Chaurasia, A. Gautam, R. Rajkumar, and A. S. Chander, Road Traffic Optimization Using Image Processing and Clustering Algorithms, Advances in Engineering Software, Vol. 181, pp. 103460, Jul. 2023, doi: 10.1016/j.advengsoft.2023.103460. V. V.-T. Tran and W.-H. Tsai, Audio-Vision Emergency Vehicle Detection, IEEE Sensors Journal, Vol. PP, No. 99, pp. 1–1, Nov. 2021, doi: 10.1109/JSEN.2021.3127893. VI. L. Sumi and V. Ranga, Intelligent Traffic Management System for Prioritizing Emergency Vehicles in a Smart City, International Journal of Engineering (IJE), Vol. 31, No. 2, pp. 278–283, Feb. 2018, doi: 10.5829/ije.2018.31.02b.11. VII. S. Abinaya, D. Cheran, C. Ranjith Kumar, P. Saravana Kumar, and others, Real-Time Road Traffic Flow Optimization Using Deep Learning Techniques, in Proc. 2025 4th OPJU Int. Technol. Conf. (OTCON) on Smart Computing for Innovation and Advancement in Industry 5.0, Apr. 2025, doi: 10.1109/OTCON65728.2025.11070592. VIII. A. M. Ashir, A Transfer-Learning-Based Approach for Emergency Vehicle Detection, Eurasian Journal of Science and Engineering, Vol. 8, No. 1, pp. 75–89, May 2022, doi: 10.23918/eajse.v8i1p75. IX. A. Ba, A. S. Ganesh, and N. Kumar, Smart Traffic Light Control System for Emergency Vehicles Using RF Module, in Futuristic Trends in Electronics & Instrumentation Engineering, Vol. 3, Book 1, pp. 208–216, Mar. 2024, doi: 10.58532/V3BDEI1P5CH3. X. K. Choudhury and D. Nandi, Detection and Prioritization of Emergency Vehicles in Intelligent Traffic Management System, in Proc. 2021 IEEE Bombay Section Signature Conference (IBSSC), Mumbai, India, Nov. 2021, doi: 10.1109/IBSSC53889.2021.9673211.