Efficient lightweight residual network for real-time road semantic segmentation

<span lang="EN-US">Intelligent transportation system (ITS) is currently one of the most discussed topics in scientific research. Actually, ITS offers advanced monitoring systems that include vehicle counting, pedestrian detection. Lately, convolutional neural networks (CNNs) are extensively used computer vision tasks, including segmentation, classification, and In fact, image semantic segmentation a critical issue applications. For example, self-driving vehicles require high accuracy with lower parameter requirements to segment road scene objects real-time. However, related work focus on side, or requirements, which make CNN models difficult use real-time order resolve this issue, we propose efficient lightweight residual network (ELRNet), novel ELRNet, an asymmetrical encoder-decoder architecture. Indeed, network, compare four varieties proposed factorized block, three loss functions get best combination. addition, model trained from scratch using only 0.61M parameters. All experiments evaluated popular public cambridge-driving labeled video database (CamVid) dataset reached results show ELRNet can achieve better performance terms parameters precision compared work.</span>