WalkGPT: Grounded Vision-Language Conversation with Depth-Aware Segmentation for Pedestrian Navigation
Paper
• 2603.10703 • Published
Official model weights for WalkGPT: Grounded Vision–Language Conversation with Depth-Aware Segmentation for Pedestrian Navigation.
📄 Paper: https://arxiv.org/abs/2603.10703
💻 Code: https://github.com/rafiibnsultan/WalkGPT
📦 Dataset: https://huggingface.co/datasets/rafiibnsultan/PAVE