GroupLane: End-to-End 3D Lane Detection with Channel-wise Grouping

This research paper introduces GroupLane, a new efficient and fast 3D lane detection system. The team at University of Hong Kong, MEGVII Technology, Tsinghua University, and Huazhong University of Science and Technology developed a method for applying row-wise classification to 3D lane detection. By leveraging both vertical and horizontal lanes recognition, they were able to create a model that outperformed the state-of-the-art PersFormer model in various metrics, including precision, inference speed, and efficiency.

Publication date: July 18, 2023
Project Page: Not available
Paper: https://arxiv.org/pdf/2307.09472.pdf