Estimation of Road Boundary for Intelligent Vehicles Based on DeepLabV3+ Architecture

Road boundary estimation is an essential task for autonomous vehicles and intelligent driving assistants. It considerably straightforward to attain the when roads are marked properly with indicators. However, estimating road reliably without prior knowledge of road, such as markings, extremely difficult. This paper proposes a method estimate boundaries in different environments deep learning-based semantic segmentation, any predefined markings. The proposed employed encoder-decoder-based DeepLab architecture segmentation types backbone networks VGG16, VGG19, ResNet-50, ResNet-101 while handling class imbalance problem by weighing loss contribution model’s outputs. performance verified using ‘ICCV09DATA’ dataset. outperformed other existing methods achieved accuracy, precision, recall, f-measure 0.9596±0.0097, 0.9453±0.0118, 0.9369±0.0149, 0.9408±0.0135 respectively RestNet-101 network Dice Coefficient function. detailed experimental analysis confirms feasibility challenging environments.