Tyre Detection System using Machine Vision

Abstract

The tyre is one of the most critical parts of a vehicle that keeps the driver safe. A prototype of tyre detection using OpenMV4 H7 plus that will notify the operators using different LED indicators for different conditions has been designed and the prototype is able to identify the tyre condition whether it is full or low pressure. The range between the tyre and the prototype is 70 cm parallelly in front of the tyre and operates when the car fully stops and no motion. Red, green and yellow LED as an indicator to notify the operators and tyre aspect ratio of 65 and diameter of the rim is 14.  OpenMV4 H7 plus is used to run the trained model, the model is trained using MobileNet V2. The MobileNet V2 has been trained using more than 300 pictures from each class of full tyre, low tyre and tyre not detected. The suitable range of the OpenMV Cam from the tyre parallelly with the camera is 70 cm so that the tyre is in the centre frame to achieve accuracy above 0.9842. The prototype is able to differentiate the tyres’ pressure with an accuracy of 0.9921, 0.9960 and 0.9960 for low pressure, full pressure and tyre not detected, respectively. The prototype has an average of 91% success rate for 20 set testing in detecting the tyre pressure. The red LED will turn on when low-pressure tyres, the green LED will on when full tyre pressure and the yellow LED will on when no tyre is detected. In this project, it can be observed that the pressure of a tyre will affect the sidewall of the tyre. As the pressure tyre decreases the sidewall decrease. A tyre pressure detection system can be used at the petrol pump. User can simply check their tyre condition from inside of their car.