A hybrid spatial–temporal deep learning architecture for lane detection

Accurate and reliable lane detection is vital for the safe performance of lane-keeping assistance departure warning systems. However, under certain challenging circumstances, it difficult to get satisfactory in accurately detecting lanes from one single image as mostly done current literature. Since markings are continuous lines, that be detected can potentially better deduced if information previous frames incorporated. This study proposes a novel hybrid spatial–temporal (ST) sequence-to-one deep learning architecture. architecture makes full use ST multiple detect very last frame. Specifically, model integrates following aspects: (a) feature extraction module equipped with spatial convolutional neural network; (b) integration constructed by recurrent (c) encoder–decoder structure, which this segmentation problem work an end-to-end supervised format. Extensive experiments reveal proposed effectively handle driving scenes outperforms available state-of-the-art methods.