Search papers, labs, and topics across Lattice.
5
6
8
0
Forget grid layouts: Map2World lets you generate consistent 3D worlds from arbitrary segment maps, offering unprecedented control and scalability.
Ditch the endless image libraries: this generative photomosaic approach uses diffusion models to create personalized, structurally coherent mosaics from just a few reference images.
Modeling human-object interactions in LiDAR data dramatically improves 3D human pose estimation, especially in cluttered scenes where interactions introduce ambiguity.
A global consensus on AI safety risks and capabilities has emerged from a panel of 100+ independent experts, representing a landmark effort in international collaboration.
Reconstructing 3D scenes just got a whole lot smarter: TeHOR uses text descriptions to understand human-object interactions beyond physical contact, leading to more accurate and contextually aware reconstructions.
