面向视觉跟踪的抗漂移时序先验
Drift-Resilient Temporal Priors for Visual Tracking
DTPTrack 提供了一种即插即用的时序可靠性校准与动态先验合成机制，可集成到现有跟踪器中抑制漂移，其 TRC 和 TGS 模块的设计思路可迁移到其他需要聚合历史帧信息的视频任务。
原文依据 we introduce DTPTrack, a lightweight and generalizable module designed to be seamlessly integrated into existing trackers to suppress drift
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论时序信息对视觉跟踪至关重要，但现有多帧跟踪器因朴素聚合带噪历史预测而容易受到模型漂移的影响。本文提出 DTPTrack，一个轻量且可泛化的模块，旨在无缝集成到现有跟踪器中以抑制漂移。我们的框架包含两个核心组件：（1）时序可靠性校准器（TRC）机制，学习为历史状态分配逐帧可靠性分数，在锚定真实模板的同时过滤噪声；（2）时序引导合成器（TGS）模块，将校准后的历史合成为一组紧凑的动态时序先验，以提供预测引导。为证明其通用性，我们将 DTPTrack 集成到三种不同的跟踪架构——OSTrack、ODTrack 和 LoRAT——中，并在所有基线上展示出一致且显著的性能提升。我们基于扩展的 LoRATv2 骨干构建的最佳模型在多个基准上创造了新的最优，在 LaSOT 上达到 77.5% 的成功率，在 GOT-10k 上达到 80.3% 的 AO。源代码见 https://github.com/NorahGreen/DTPTrack。
英文原文摘要 1203 字符
Temporal information is crucial for visual tracking, but existing multi-frame trackers are vulnerable to model drift caused by naively aggregating noisy historical predictions. In this paper, we introduce DTPTrack, a lightweight and generalizable module designed to be seamlessly integrated into existing trackers to suppress drift. Our framework consists of two core components: (1) a Temporal Reliability Calibrator (TRC) mechanism that learns to assign a per-frame reliability score to historical states, filtering out noise while anchoring on the ground-truth template; and (2) a Temporal Guidance Synthesizer (TGS) module that synthesizes this calibrated history into a compact set of dynamic temporal priors to provide predictive guidance. To demonstrate its versatility, we integrate DTPTrack into three diverse tracking architectures--OSTrack, ODTrack, and LoRAT--and show consistent, significant performance gains across all baselines. Our best-performing model, built upon an extended LoRATv2 backbone, sets a new state-of-the-art on several benchmarks, achieving a 77.5% Success rate on LaSOT and an 80.3% AO on GOT-10k. The source code is available at https://github.com/NorahGreen/DTPTrack.