Channel Attention in LiDAR-camera Fusion for Lane Line Segmentation

作者：

Highlights：

• A channel attention mechanism is proposed to improve LiDAR-camera fusion in lane line segmentation;

• The CFECA(Cross Fusion Efficient Channel Attention) module is designed and applied to improve the fusion method and allow abundant LiDAR-camera fusion information to be used simultaneously across channels;

• A method for determining fusion weights, which are transferable in multimodal fusion, is proposed.

摘要

•A channel attention mechanism is proposed to improve LiDAR-camera fusion in lane line segmentation;•The CFECA(Cross Fusion Efficient Channel Attention) module is designed and applied to improve the fusion method and allow abundant LiDAR-camera fusion information to be used simultaneously across channels;•A method for determining fusion weights, which are transferable in multimodal fusion, is proposed.

论文关键词：LiDAR-camera fusion,Lane line segmentation,Channel attention mechanism,Multimodal fusion,Fusion information

论文评审过程：Received 12 August 2020, Revised 26 March 2021, Accepted 4 May 2021, Available online 15 May 2021, Version of Record 29 May 2021.

论文官网地址：https://doi.org/10.1016/j.patcog.2021.108020