实例级视觉主动跟踪与遮挡感知规划
Instance-level Visual Active Tracking with Occlusion-Aware Planning
提出实例级原型初始化与在线增强机制，结合置信度感知卡尔曼滤波提升相似干扰物下的判别力；并利用条件扩散生成避障轨迹实现遮挡恢复，为视觉跟踪中的主动规划提供可复用范式。
原文依据 uses conditional diffusion to generate obstacle-avoiding paths for occlusion recovery
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觉主动跟踪（VAT）旨在控制相机在三维空间中跟随目标，对无人机导航和安全监控等应用至关重要。然而，在实际部署中面临两个关键瓶颈：实例级判别不足导致视觉相似干扰物造成混淆，以及缺乏主动规划导致遮挡下严重失败。为解决这些问题，我们提出OA-VAT，一个包含三个互补模块的统一流水线。首先，无需训练的实例感知离线原型初始化通过DINOv3聚合多视角增强特征，构建具有判别力的实例原型，缓解干扰物混淆。其次，在线原型增强跟踪器在线增强原型，并集成置信度感知的卡尔曼滤波器，在外观和运动变化下实现稳定跟踪。第三，遮挡感知轨迹规划器在我们新构建的Planning-20k数据集上训练，使用条件扩散生成避障路径以恢复遮挡。实验表明，OA-VAT在UnrealCV上达到0.93的平均成功率（相比SOTA TrackVLA提升2.2%），在真实世界数据集上达到90.8%的平均CAR（相比SOTA GC-VAT提升12.1%），在DJI Tello无人机上达到81.6%的TSR。在RTX 3090上以35 FPS运行，为实际部署提供鲁棒、实时的性能。代码见https://github.com/SHWplus/OA-VAT。
英文原文摘要 1413 字符
Visual Active Tracking (VAT) aims to control cameras to follow a target in 3D space, which is critical for applications like drone navigation and security surveillance. However, it faces two key bottlenecks in real-world deployment: confusion from visually similar distractors caused by insufficient instance-level discrimination and severe failure under occlusions due to the absence of active planning. To address these, we propose OA-VAT, a unified pipeline with three complementary modules. First, a training-free Instance-Aware Offline Prototype Initialization aggregates multi-view augmented features via DINOv3 to construct discriminative instance prototypes, mitigating distractor confusion. Second, an Online Prototype Enhancement Tracker enhances prototypes online and integrates a confidence-aware Kalman filter for stable tracking under appearance and motion changes. Third, an Occlusion-Aware Trajectory Planner, trained on our new Planning-20k dataset, uses conditional diffusion to generate obstacle-avoiding paths for occlusion recovery. Experiments demonstrate OA-VAT achieves 0.93 average SR on UnrealCV (+2.2% vs. SOTA TrackVLA), 90.8% average CAR on real-world datasets (+12.1% vs. SOTA GC-VAT), and 81.6% TSR on a DJI Tello drone. Running at 35 FPS on an RTX 3090, it delivers robust, real-time performance for practical deployment. The code is available at https://github.com/SHWplus/OA-VAT.