Ambulance Detection System

This repository contains code and resources for a cloud-based ambulance detection system using YOLOv8.

Overview

The project aims to minimize ambulance response time to emergency calls by detecting ambulance vehicles and synchronizing them with traffic cameras and signaling systems. The system utilizes machine learning techniques, specifically YOLOv8, for real-time ambulance detection.

Key Features

YOLOv8 Model: The core of the system is the YOLOv8 model, trained to detect ambulance vehicles in real-time. Dataset: A dataset comprising 3000 images of ambulance vehicles from 10 different countries is included for training and evaluation purposes. Layered Architecture: The system architecture consists of multiple layers, including the data acquisition layer (DAL), ambulance detection layer (ADL), monitoring layer (ML), and cloud layer (CL), to support cloud-based ambulance detection.

Data

Country	Data Type	Total Images	Total Labels
Ambulance - Germany	train	200	220
Ambulance - Germany	val	100	112
Ambulance - Italy	train	200	203
Ambulance - Italy	val	100	111
Ambulance - Japan	train	200	211
Ambulance - Japan	val	100	113
Ambulance - Norway	train	200	219
Ambulance - Norway	val	100	108
Ambulance - Russia	train	200	253
Ambulance - Russia	val	100	116
Ambulance - Saudi Arabia	train	200	215
Ambulance - Saudi Arabia	val	100	104
Ambulance - Spain	train	200	233
Ambulance - Spain	val	100	114
Ambulance - Sweden	train	200	216
Ambulance - Sweden	val	100	112
Ambulance - Turkey	train	200	250
Ambulance - Turkey	val	100	111
Ambulance - United Kingdom	train	200	286
Ambulance - United Kingdom	val	100	131

Models

Country	Model Type	Accuracy -
Germany	Pre-trained	0.985
Germany	Custom	0.956
Italy	Pre-trained	0.945
Italy	Custom	0.903
Japan	Pre-trained	0.995
Japan	Custom	0.938
Norway	Pre-trained	0.942
Norway	Custom	0.895
Russia	Pre-trained	0.993
Russia	Custom	0.957
Saudi Arabia	Pre-trained	0.991
Saudi Arabia	Custom	0.98
Spain	Pre-trained	0.969
Spain	Custom	0.958
Sweden	Pre-trained	0.994
Sweden	Custom	0.976
Turkey	Pre-trained	0.988
Turkey	Custom	0.944
United Kingdom	Pre-trained	0.987
United Kingdom	Custom	0.972
Universal - Yolov7	Pre-trained	0.953
Universal - Yolov7	Custom	0.953
Universal - Yolov5	Pre-trained	0.979
Universal - Yolov5	Custom	0.979
Universal - Yolov8	Pre-trained	0.982
Universal - Yolov8	Custom	0.982

contact

Access to Data and Models: For access to the dataset and the YOLOv8 model provided in the dataset and model directories respectively, please contact us at: ziadalgurafi@gmail.com, tnoor@taibahu.edu.sa, anoor@taibahu.edu.sa.

Deploy the system and integrate it with traffic cameras and signaling systems for real-time ambulance detection. For any assistance regarding the system, please contact us at basil.khalid.alharbi@gmail.com.