ChronoVision：基于潜在状态重建的时间推理

摘要

多模态大语言模型在被动感知方面表现出色，但在需要多步时序推理的复杂视觉认知任务中却面临困难。这种性能下降主要源于基于语言的推理固有一定的模糊性，难以准确描述连续的视觉变换。为解决这一问题，我们提出ChronoVision，一种旨在将视觉逻辑与潜在图像表征对齐的多模态框架。在监督微调阶段，重建视觉头（Reconstructive Visual Head）预测最终变换状态的潜在表示，而感兴趣区域注意力定位模块（ROI Attention Locating module）通过语义区间查询使模型聚焦于关键视觉证据。在后训练阶段，我们应用带有隐式过程对齐机制的强化学习，由复合奖励函数引导，该函数综合评估结果正确性、潜在过程对齐度以及无监督视觉聚焦度。此外，我们引入了Vbvr-VQA，一个新颖的数据集，通过将视频推理重构为严格的图像排序任务来评估时序跟踪能力。实验表明，ChronoVision在Vbvr-VQA上取得了最先进的性能，域内准确率达74.8%，域外准确率达71.6%，并在极具挑战性的跨域基准IntPhys2上取得55.0%的强劲准确率。

English

Multimodal large language models excel at passive perception but struggle with complex visual cognitive tasks requiring multi-step temporal reasoning. This degradation largely stems from the inherent ambiguity of language-based reasoning, which often fails to accurately articulate continuous visual transformations. To address this, we propose ChronoVision, a multimodal framework designed to align visual logic with latent imagery. During supervised fine-tuning, a Reconstructive Visual Head predicts the latent representation of the final transformed state, while an ROI Attention Locating module focuses the model on key visual evidence via semantic span queries. In post-training, we apply reinforcement learning with an implicit process grounding mechanism, guided by a composite reward function that evaluates outcome correctness, latent process alignment, and unsupervised visual focus. Furthermore, we introduce Vbvr-VQA, a novel dataset that evaluates temporal tracking by reformulating video reasoning into a strict image-ordering task. Experiments demonstrate that ChronoVision achieves state-of-the-art performance on Vbvr-VQA with 74.8% in-domain and 71.6% out-of-domain accuracy, alongside a strong 55.0% accuracy on IntPhys2, a highly challenging cross-domain benchmark.