VRR-QA：超越显式线索的视频视觉关系推理
VRR-QA: Visual Relational Reasoning in Videos Beyond Explicit Cues
为视频问答提供了专门考察跨帧隐式关系推理的评测基准，可用于诊断现有 VideoQA 模型对表层视觉线索的依赖，并作为隐式推理能力训练与评估的测试平台。
原文依据 we introduce VRR-QA, a benchmark for Visual Relational Reasoning Beyond Explicit Cues
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论视频问答（VideoQA）通过多模态学习对齐视觉与文本模态已取得显著进展。然而，现有基准绝大多数聚焦于可通过显式视觉内容——即在单帧或短片段中可直接观察到的动作、物体与事件——回答的问题。要像人类一样真正理解视频，模型必须超越直接呈现的内容，推断仅在跨帧之间隐含的隐藏关系与上下文线索。现有基准未能捕捉视频理解的这一关键方面。为填补这一空白，我们提出 VRR-QA，一个面向超越显式线索的视觉关系推理的基准。我们从电影等创意与影视化视频中构建该基准，这类视频刻意运用叙事手法省略对某些事件或关系的直接描绘，要求观众自行推断。VRR-QA 包含 1K 条由专家精心标注的问答对，取自 1K 个创意视频片段，覆盖 15 种类型、横跨 70 年的内容，涵盖真人实拍与动画作品。我们对 14 个领先 VideoQA 模型的广泛评测揭示了持续且显著的性能下降，凸显其对表层视觉线索的依赖，并表明隐式推理的难度。即便最佳模型也仅达 64% 准确率，大幅低于人类基线。模型间的性能差异进一步体现了 VRR-QA 所呈现挑战的复杂性与多样性。通过发布数据集与数据收集框架，VRR-QA 为推进 VideoQA 建立了严谨、多样且可复现的测试平台。
英文原文摘要 1687 字符
Video Question Answering (VideoQA) has made significant strides by leveraging multimodal learning to align visual and textual modalities. However, current benchmarks overwhelmingly focus on questions answerable through explicit visual content - actions, objects, and events - directly observable within individual frames or short clips. To truly understand videos as humans do, models must go beyond what is directly shown, inferring hidden relationships and contextual cues that are only implied across frames. Current benchmarks fail to capture this essential aspect of video understanding. To address this gap, we introduce VRR-QA, a benchmark for Visual Relational Reasoning Beyond Explicit Cues. We curate our benchmark from creative and cinematic videos such as movies, that deliberately employ storytelling techniques which omit direct depictions of certain events or relations, requiring viewers to infer them. VRR-QA comprises 1K meticulously expert-annotated QA pairs drawn from 1K creative video clips covering 15 genres across 7 decades of content, from both live-action and animated titles. Our extensive evaluations on 14 leading VideoQA models reveals consistent and significant performance degradation, underscoring their reliance on surface-level visual cues and highlighting the difficulty of implicit reasoning. Even the best model substantially underperforms human baselines with only 64% accuracy. Performance variations across models further illustrate the complexity and diversity of the challenges presented by VRR-QA. By releasing both dataset and data collection framework, VRR-QA establishes a rigorous, diverse, and reproducible testbed for advancing VideoQA.