Real-Time Cell Phone Recognition for Motorcycle Emission Inspection System

The impacts of air pollution on public health are becoming increasingly important. Taiwan has the highest proportion of mobile sources of pollution, which include automobiles, motorcycles, ships, and airplanes. These oil-fueled vehicles all emit pollutants that are harmful to humans, animals, plants, and the environment. The most direct way to reduce mobile pollution is through electric vehicles. The next approach is to require vehicles that have been in use for over five years to undergo regular emission inspections. However, many drivers often forget to perform regular emission inspections. Therefore, we propose an effective solution to improve vehicle emission inspection rates. Our system consists of three main components, the mobile application, the server-side system and the web interface. The system uses YOLOv8 on the mobile application for license plate recognition and STAR-Net to confirm license plate numbers. A web crawler is then employed to check whether the vehicle has undergone its regular emission inspection. The Cell phone test YOLOv8 model for license plate recognition with 95% accuracy, execution speed up to 22 to 30fps. The license plate numbers with 92% recognition rate using STAR-Net. The server-side system takes only 7 seconds from data reception to the completion of web crawling.