#visual correspondence

Meta AI presents Multi-SpatialMLLM, a new model improving multi-frame spatial understanding in multi-modal large language models, supported by the extensive MultiSPA dataset and benchmark.