Semantic segmentation of autonomous driving scenes based on multi-scale adaptive attention mechanism

Frontiers in Neuroscience

Neuroscience

Lei Wang¹  Jun Wang¹  Danping Liu¹  Dong Zhang² 

[1] School of Advanced Manufacturing Engineering, Hefei University, Hefei, China;State Key Laboratory of Automotive Simulation and Control, Jilin University, Changchun, China;

关键词: semantic segmentation;    attention mechanism;    autonomous driving;    convolutional neural networks;    deep learning;   

DOI  :  10.3389/fnins.2023.1291674

 received in 2023-09-09, accepted in 2023-10-06,  发布年份 2023

来源: Frontiers

【 摘 要 】

IntroductionSemantic segmentation is a crucial visual representation learning task for autonomous driving systems, as it enables the perception of surrounding objects and road conditions to ensure safe and efficient navigation.MethodsIn this paper, we present a novel semantic segmentation approach for autonomous driving scenes using a Multi-Scale Adaptive Mechanism (MSAAM). The proposed method addresses the challenges associated with complex driving environments, including large-scale variations, occlusions, and diverse object appearances. Our MSAAM integrates multiple scale features and adaptively selects the most relevant features for precise segmentation. We introduce a novel attention module that incorporates spatial, channel-wise and scale-wise attention mechanisms to effectively enhance the discriminative power of features.ResultsThe experimental results of the model on key objectives in the Cityscapes dataset are: ClassAvg:81.13, mIoU:71.46. The experimental results on comprehensive evaluation metrics are: AUROC:98.79, AP:68.46, FPR95:5.72. The experimental results in terms of computational cost are: GFLOPs:2117.01, Infer. Time (ms):61.06. All experimental results data are superior to the comparative method model.DiscussionThe proposed method achieves superior performance compared to state-of-the-art techniques on several benchmark datasets demonstrating its efficacy in addressing the challenges of autonomous driving scene understanding.

【 授权许可】

Unknown   
Copyright © 2023 Liu, Zhang, Wang and Wang.