roadsegmentation

AI-powered road resilience system that detects and reconstructs occluded roads from satellite imagery using deep learning, enabling accurate mapping for disaster response, infrastructure planning, and smart mobility.

A deep learning-based road segmentation model for autonomous vehicles, MESNet integrates ResNet-50, VGG-16, and PSPNet to achieve high accuracy and precision in diverse environments. Trained on the KITTI dataset, it handles real-time road segmentation tasks with robust performance metrics.

A framework for enhancing noisy ground truths in network-like infrastructure by leveraging domain-specific graph constraints and optimization properties to improve overall segmentation results.