Video generation models are increasingly being explored as world simulators for embodied planning and learning. To do so effectively, these models must not only generate visually appealing frames, but also predict how environments dynamically evolve when executing goal-directed actions. While evaluating these capabilit...
Patrick Amadeus Irawan, Iskandar Muda Rizky Parlambang, Rava Maulana et al.· 0 citations
Multimodal large language models (MLLMs) struggle with spatial, embodied, physical, and temporal reasoning. We hypothesize that these failures reflect a shared deficit in visual transition reasoning, and test whether this capability can serve as a shared training primitive, one that different models can learn from diff...
Zhe-Yu Fan, Yue Zhang, Ming-Kai Deng et al.· 0 citations
This work introduces LAGO (LAnguage-Guided adaptive Object-region focus), which reframes localized recognition as language-guided directed region discovery and addresses the circular dependence between recognizing a class and locating its supporting evidence, while preserving complementary local, contextual, and global...
Jun-Yi Hu, Qiji Zhou, Lei Zhang et al.· arXiv.org· 0 citations
AnchorWeave is introduced, a memory-augmented video generation framework that replaces a single misaligned global memory with multiple clean local geometric memories and learns to reconcile their cross-view inconsistencies.
Zun Wang, Han Lin, Jaehong Yoon et al.· arXiv.org· 18 citations
V-Co, a systematic study of visual co-denoising in a unified JiT-based framework, outperforms the underlying pixel-space diffusion baseline and strong prior pixel-diffusion methods while using fewer training epochs, offering practical guidance for future representation-aligned generative models.
Han Lin, Xichen Pan, Zun Wang et al.· arXiv.org· 0 citations
This work introduces DEER-3D, an error-driven framework that diagnoses grounding failures and generates targeted counterfactual training supervision via a structured "Decompose, Diagnose, Edit, and Retrain"loop, and underscores the effectiveness of targeted, error-driven scene editing in bridging linguistic reasoning w...
Yue Zhang, Zun Wang, Han Lin et al.· arXiv.org· 0 citations
EgoMemReason is introduced, a comprehensive benchmark for week-long egocentric video understanding through memory-driven reasoning that evaluates three complementary memory types: entity memory, tracking how object states evolve and change across days; event memory, recalling and ordering activities separated by hours...
