Fumes and Smoke Car Detection Using YOLOv8

Abstract

Reducing vehicle emissions is crucial for achieving environmental sustainability and mitigating the detrimental impact of air pollution. This scientific article explores the utilization of the YOLOv8 object detection algorithm to detect and mitigate smoke emissions from cars, contributing to the overall reduction of vehicle emissions. By training the YOLOv8 model using a dataset obtained from RoboFlow, consisting of annotated images of smoke-emitting cars, a robust and accurate detection system is developed. The article delves into the process of dataset preparation, including the annotation of images and the creation of corresponding text files with bounding box coordinates around smoke-emitting vehicles. Through iterative model training, which involves optimizing the model's parameters and loss function using backpropagation, the YOLOv8 model becomes adept at detecting and classifying smoke-emitting cars. Once trained, the model can be deployed in real-time scenarios to identify vehicles emitting excessive smoke. Upon detection, measures can be implemented to minimize the smoke percentage in the air. These actions may encompass notifying the driver, suggesting maintenance actions, or even automatically adjusting factors such as vehicle speed and fuel mixture. The integration of fumes and smoke car detection using YOLOv8 into sustainable technologies offers several advantages, including proactive identification of smoke-emitting vehicles and targeted interventions to reduce emissions. This approach fosters responsible vehicle maintenance practices and encourages eco-friendly behaviors among drivers, ultimately contributing to a greener and healthier environment. By embracing these advancements and reducing the smoke percentage in the air, we can take significant strides towards a more sustainable future for all.