通过推理生成描述，实现精准视频编辑检索。
Reason-Then-Retrieve for CoVR-R with Structured Edit Prompts and Dense-Sparse Fusion
- 先推理编辑指令，再生成目标视频描述作为查询向量。
- 在测试集上达到R@1 89.73，R@50 97.98的高精度。
- 适合需要细粒度视频修改与检索的研究者。
CoVR-R研究的是基于推理的复合视频检索：给定参考视频和编辑指令，系统需检索出满足编辑要求的目标视频。难点在于目标视频未被直接描述，需从物体身份、动作顺序、最终状态、手部交互和场景转换等细微变化中推断。本文构建了基于Qwen3.5-27B的零样本“推理-检索”流程。对每个候选视频，模型生成结构化描述并用加权池化隐藏状态得到稠密嵌入；对每个查询，模型先在参考视频和指令上进行编辑推理，再生成目标视频描述，其隐藏状态作为查询嵌入。同时引入基于生成文本的TF-IDF分支，通过分任务权重融合两种排序结果。验证集上最佳提交达R@1 80.81，R@5 94.86，R@10 97.11，R@50 98.59；盲测集上达R@1 89.73，R@5 95.79，R@10 96.63，R@50 97.98。
原文摘要 · Abstract (English)
CoVR-R studies reason-aware composed video retrieval: given a reference video and an edit instruction, the system must retrieve the target video that satisfies the edit. The main difficulty is that the target is not described directly; it must be inferred from fine-grained changes in object identity, action order, final state, hand interaction, and scene transition. We build a zero-shot reason-then-retrieve pipeline around Qwen3.5-27B. For each gallery video, the model generates a retrieval-oriented structured description and a dense embedding by pooling generated-token hidden states with token-dependent weights. For each query, the model first performs edit reasoning over the reference video and instruction, then generates a target-video description whose hidden states serve as the query embedding. We complement dense retrieval with a TF-IDF branch over the generated texts and fuse the two rankings with split-specific weights. On validation, the current best submission reaches 80.81 at R@1, 94.86 at R@5, 97.11 at R@10, and 98.59 at R@50. On the blind test split, it reaches 89.73 at R@1, 95.79 at R@5, 96.63 at R@10, and 97.98 at R@50.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。