跟随显著性：面向检索增强密集视频描述的有监督显著性
Follow the Saliency: Supervised Saliency for Retrieval-augmented Dense Video Captioning
把 DVC 真值直接转成二值标签训练帧级显著性/高光检测器，再用同一显著性分数同时驱动检索分段与解码器 Saliency Prompt，为视频-语言检索增强生成提供了一个无需额外标注的统一时序信号。
原文依据 supervising frame-level saliency through a highlight detection module
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论现有的检索增强密集视频描述（DVC）方法往往无法实现与真实事件边界对齐的准确时序分段，因为它们依赖忽略真值事件边界的启发式策略。所提出的 STaRC 框架通过一个高光检测模块对帧级显著性进行监督，从而克服了这一局限。值得注意的是，该高光检测模块使用直接由 DVC 真值标注导出的二值标签进行训练，无需额外标注。我们还提出将显著性分数用作统一的时序信号：它通过显著性引导的分段来驱动检索，并通过注入解码器的显式显著性提示（Saliency Prompt）来指导描述生成。通过施加显著性约束的分段，本方法产生与真实事件转换紧密对齐的时序连贯片段，从而实现更准确的检索和具有上下文依据的描述生成。我们在 YouCook2 和 ViTT 基准上进行了全面评测，STaRC 在大多数指标上取得了最优性能。
英文原文摘要 1137 字符
Existing retrieval-augmented approaches for Dense Video Captioning (DVC) often fail to achieve accurate temporal segmentation aligned with true event boundaries, as they rely on heuristic strategies that overlook ground truth event boundaries.The proposed framework, STaRC, overcomes this limitation by supervising frame-level saliency through a highlight detection module. Note that the highlight detection module is trained on binary labels derived directly from DVC ground truth annotations without the need for additional annotation.We also propose to utilize the saliency scoresas a unified temporal signal that drives retrieval via saliency-guided segmentation and informs caption generation through explicit Saliency Prompts injected into the decoder.By enforcing saliency-constrained segmentation, our method produces temporally coherent segments that align closely with actual event transitions, leading to more accurate retrieval and contextually grounded caption generation. We conduct comprehensive evaluations on the YouCook2 and ViTT benchmarks, where STaRC achieves state-of-the-art performance across most of the metrics.