Tracking Is Not Permanence: What Video World Models Keep of a Hidden Object
cs.CV, cs.CL
Submitted: 2026-10-05
Updated: 2026-10-05
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Terminology
Sources
- V-JEPA 2: Self-Supervised Video Models Enable Understanding, Prediction and Planning
- Qwen3-VL Technical Report
- Unifying (Machine) Vision via Counterfactual World Modeling
- IntPhys 2: Benchmarking Intuitive Physics Understanding In Complex Synthetic Environments
- The Invisible Hand of Physics: When Video Diffusion Models Know More Than They Show
- Intuitive physics understanding emerges from self-supervised pretraining on natural videos
- EgoDex: Learning Dexterous Manipulation from Large-Scale Egocentric Video
- Interpreting Physics in Video World Models
- Current World Models Lack a Persistent State Core
- Do generative video models understand physical principles?
- V-JEPA 2.1: Unlocking Dense Features in Video Self-Supervised Learning
- Cosmos World Foundation Model Platform for Physical AI
- Long-Context State-Space Video World Models
- How Do Video Foundation Models Encode Intuitive Physics? Probing Across Pretraining Paradigms
- Occlusion resistant learning of intuitive physics from videos
- Can Video World Models Track Unobserved World States?
- WorldMem: Long-term Consistent World Simulation with Memory
- HERA: Historical Evidence Routing Adapter for Physical Prediction in Latent World Models
- MediaPipe Hands: On-device Real-time Hand Tracking
Related papers
- Loss Knows Best: Detecting Annotation Errors in Videos via Loss Trajectories
- AnchorWeave: World-Consistent Video Generation with Retrieved Local Spatial Memories
- Benchmarking the Robustness of Foundation Models for Mammography under Domain Shift
- MambaX-Net: Dual-Input Mamba-Enhanced Cross-Attention Network for Longitudinal MRI Segmentation
- TeleOCR: Navigating Document Parsing Across Digital and Camera-Captured Documents
- A Survey on Efficient Vision-Language-Action Models