Benchmarking VLMs on Cross-View Camera Orientation Estimation
Developed a cross-view camera orientation estimation benchmark for vision-language models to evaluate spatial reasoning across aerial and street-view imagery for GPS-denied navigation.
← All projects