UrbanFlow-AI Traffic Management
AI-powered real-time traffic signal optimization using YOLOv11 and AWS services.
Duration
4 weeks
Team
2 members
Accuracy
94.6%
Technologies Used
Project Overview
UrbanFlow uses YOLOv11 to detect vehicle types and counts in real-time video feeds. The data is processed and stored using AWS (Lambda, S3, DynamoDB) to dynamically manage traffic signal durations based on road congestion, improving urban mobility.
Methodology
Challenges
Handling real-time object detection across four camera angles
Deploying ML models with low latency on the cloud
Designing a scalable, event-driven architecture
Solutions
Trained YOLOv11 model with high mAP@50 (94.6%) for multiple vehicle classes
Used AWS Lambda + S3 triggers for near real-time model inference
Leveraged DynamoDB to log counts and compute congestion trends
Results & Impact
Successfully identified cars, trucks, buses, bikes, and pedestrians
Reduced signal wait time on average by 22% in test scenarios
Scalable architecture that supports multiple junctions
Interested in This Project?
Want to learn more about the implementation details or discuss similar projects? I'd love to hear from you!