YOLOv11-based emergency vehicle detection

As the number of cars on the roads are rising, we often see the issue of traffic congestion, which makes it difficult for emergency vehicles to move and can result in life or death situations. In order to address this issue, a YOLOv11 based emergency vehicle detection model was created, which is trained to recognise the lights of emergency vehicles. A custom labeled dataset with different lighting, weather, and urban scenarios was used to ensure effectiveness of model. The model gone through training for over 320 epochs, achieving a mean Average Precision of approximately 78% and mAP@50:95 of approximately 68%, with high precision 89.8% and acceptable recall 65.8%. The methodology included preprocessing, inference at 37 FPS, and performance visualization through bounding boxes and confidence scores. Tests confirmed practical utility, showcasing 98% accuracy in detecting ambulance lights, making the system as a good solution for emergency response.