Search papers, labs, and topics across Lattice.
1
3
LVLMs can achieve state-of-the-art performance in fine-grained visual reasoning tasks by mimicking human coarse-to-fine cognitive processing with a hierarchical architecture.
