鲁棒的提示式视频目标分割
Robust Promptable Video Object Segmentation
把「记忆中的目标级表征」作为条件去门控低秩适配（LoRA 式）参数，使鲁棒化过程按被跟踪目标分别进行并保持时序一致；这套 memory-conditioned gated adaptation 可直接迁移到其他提示式视频模型（如 SAM2 类跟踪器）做测试时/轻量鲁棒适配。
原文依据 MoGA leverages object-specific representations maintained in memory across frames to condition the robustification process
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论提示式视频目标分割（PVOS）模型在输入损坏下的性能会显著下降，这阻碍了 PVOS 在安全关键领域的部署。本文首次对鲁棒 PVOS（RobustPVOS）进行了系统性研究。我们首先构建了一个新的综合性基准，包含两个真实世界评测数据集，涵盖 351 段视频片段和 2,500 余个目标掩码，均处于真实世界的恶劣条件下。同时，我们对现有 VOS 数据集施加多样且随时间变化的损坏，从而生成合成训练数据。此外，我们提出了一种新的 RobustPVOS 方法，称为记忆-目标条件门控秩适配（Memory-object-conditioned Gated-rank Adaptation，MoGA）。成功实现 RobustPVOS 的关键有两点：有效处理目标特定的退化，以及确保预测的时序一致性。MoGA 利用跨帧保存在记忆中的目标特定表征来对鲁棒化过程进行条件化，使模型能够以时序一致的方式对每个被跟踪目标区别处理。在我们基准上的大量实验验证了 MoGA 的有效性，在合成与真实世界数据集上、在多种损坏类型下均取得一致且显著的提升，为未来 RobustPVOS 研究建立了强有力的基线。我们的基准已公开于 https://sohyun-l.github.io/RobustPVOS_project_page/。
英文原文摘要 1398 字符
The performance of promptable video object segmentation (PVOS) models substantially degrades under input corruptions, which prevents PVOS deployment in safety-critical domains. This paper offers the first comprehensive study on robust PVOS (RobustPVOS). We first construct a new, comprehensive benchmark with two real-world evaluation datasets of 351 video clips and more than 2,500 object masks under real-world adverse conditions. At the same time, we generate synthetic training data by applying diverse and temporally varying corruptions to existing VOS datasets. Moreover, we present a new RobustPVOS method, dubbed Memory-object-conditioned Gated-rank Adaptation (MoGA). The key to successfully performing RobustPVOS is two-fold: effectively handling object-specific degradation and ensuring temporal consistency in predictions. MoGA leverages object-specific representations maintained in memory across frames to condition the robustification process, which allows the model to handle each tracked object differently in a temporally consistent way. Extensive experiments on our benchmark validate MoGA's efficacy, showing consistent and significant improvements across diverse corruption types on both synthetic and real-world datasets, establishing a strong baseline for future RobustPVOS research. Our benchmark is publicly available at https://sohyun-l.github.io/RobustPVOS_project_page/.