Search papers, labs, and topics across Lattice.
Affiliation:
6
0
7
Continuous video understanding doesn't require constant multimodal LLM compute: selectively waking a heavyweight reasoner via lightweight spatial filtering delivers real-time, causal anomaly reasoning without diluting rare events.
Online 3D reconstruction failure on long videos is not a representation collapse but an artifact of single-anchor pose extrapolation鈥攁nd querying relative poses across multiple keyframes fixes it with just 1% parameter overhead.
PhysCaP enables robots to actively infer hidden physical properties, achieving superior manipulation efficiency without additional sensors.
Grounding image dehazing in the physics of haze formation allows HNDiff to achieve unprecedented restoration quality, outperforming traditional methods.
DDTNet reveals that focusing on degradation transfer can dramatically improve image restoration performance across diverse weather conditions.
SpatialClaw enables agents to dynamically compose and adapt their reasoning strategies, achieving a remarkable 11.2-point accuracy boost over traditional spatial agents.
