Implicit Geometry Representations for Vision-and-Language Navigation from Web Videos

Mingfei Han, Haihong Hao, Liang Ma +6

Vision-and-Language Navigation (VLN) has long been constrained by the limited diversity and scalability of simulator-curated datasets, which fail to capture the complexity of real-…