Search papers, labs, and topics across Lattice.
5
0
7
0
DPC-Net achieves unprecedented image restoration quality by seamlessly integrating semantic understanding with low-level visual cues, outperforming existing methods on multiple benchmarks.
Adaptive Org Routing outperforms fixed collaboration protocols by dynamically selecting the best approach for each task, revealing that organizational design must be revalidated for each model family.
By decoupling geometry and texture cues, GT-PCQA overcomes the texture bias of MLLMs to achieve state-of-the-art point cloud quality assessment, even with limited PCQA data.
Transferring image quality knowledge to point clouds can dramatically improve no-reference point cloud quality assessment, but only if you carefully align features by quality level and augment features in a quality-aware way.
The YT-NTU-AVQ dataset, 10x larger than previous AVQA datasets, unlocks new possibilities for training and evaluating multimodal perception models by offering unprecedented scale and diversity.
