PolyLaneDet: Lane Detection with Free-Form Polyline

다국어 초록 (Multilingual Abstract)

Lane detection is a critical component of autonomous driving technologies that face challengessuch as varied road conditions and diverse lane orientations. In this study, we aim to addressthese challenges by proposing PolyLaneDet, a novel lane detection model that utilizes a freeform polyline, termed ‘polylane,’ which adapts to both vertical and horizontal lane orientationswithout the need for post-processing. Our method builds on the YOLO v4 architecture toavoid restricting the number of detectable lanes. This model can regress both vertical andhorizontal coordinates, thereby improving the adaptability and accuracy of lane detection invarious scenarios. We conducted extensive experiments using the CULane benchmark and acustom dataset to validate the effectiveness of the proposed approach. The results demonstratethat PolyLaneDet achieves a competitive performance, particularly in detecting horizontallane markings and stop lines, which are often omitted in traditional models. In conclusion,PolyLaneDet advances lane detection technology by combining flexible lane representationwith robust detection capabilities, making it suitable for real-world applications with diverseroad geometries.