基于双注意力机制的街景语义分割

Semantic Segmentation of Street Scenes Based on Double Attention Mechanism

高性能语义分割算法由于自身高延迟性存在无法快速感知路况的问题。本文提出一种基于注意力机制的双路径网络模型。该网络模型采用轻量的局部轮廓信息提取模块和语义信息提取模块来替代复杂的编码器结构。针对不同路径下特征图的特点,分别基于自注意力和通道注意力机制设计特征优化模块,该算法可有效地提高轻量网络结构对细节特征的表达能力。设计的语义分割网络以25 fps的速度处理图像的同时,可保持73.9%的平均交并比。经实物验证,表明本文算法具备实时性,具有一定的实际应用价值。

High-performance semantic segmentation algorithms cannot quickly perceive road conditions due to their high latency.This paper proposes a dual-path network model based on attention mechanism.The network model uses a lightweight local contour information extraction module and a semantic information extraction module to replace the complex encoder structure.Aiming at the characteristics of feature maps under different paths,feature optimization modules are designed based on self-attention and channel attention mechanisms.This algorithm effectively improves the ability of lightweight network structures to express detailed features.The designed semantic segmentation network processes images at a speed of 25 fps while maintaining an average cross-to-parallel ratio of 73.9%.The physical verification shows that the algorithm has real-time performance and high value in certain practical application.