Lane Detection of Smart Car based on Deep Learning

Abstract

Abstract Deep learning brings higher accuracy to lane detection, and the speed of the model is becoming faster and faster. The lane detection is regarded as the classification task after image gridding, and a simplified lane structure loss function is proposed, which is more suitable for the car track. The lightweight SqueezeNet is used as the backbone network, and the deep learning model is applied to the intelligent car of raspberry pie. Through the experiment, the real-time detection speed reaches 23FPS, which can complete the task of lane tracking. Compared with traditional methods, deep learning model has better robustness.