Multi-SpatialMLLM integrates depth perception, visual correspondence, and dynamic perception into MLLMs via a 27M-sample MultiSPA dataset and benchmark, yielding gains on multi-frame spatial tasks.
Tapvid-3d: A benchmark for tracking any point in 3d
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
fields
cs.CV 2verdicts
UNVERDICTED 2representative citing papers
GenMatter is a generative hierarchical model that groups low-level motion and high-level features into particles and clusters representing independently moveable physical entities, validated across dot kinematograms, camouflaged objects, and RGB videos.
citing papers explorer
-
Multi-SpatialMLLM: Multi-Frame Spatial Understanding with Multi-Modal Large Language Models
Multi-SpatialMLLM integrates depth perception, visual correspondence, and dynamic perception into MLLMs via a 27M-sample MultiSPA dataset and benchmark, yielding gains on multi-frame spatial tasks.
-
GenMatter: Perceiving Physical Objects with Generative Matter Models
GenMatter is a generative hierarchical model that groups low-level motion and high-level features into particles and clusters representing independently moveable physical entities, validated across dot kinematograms, camouflaged objects, and RGB videos.