发表机构
Beihang University; Peking University（北京航空航天大学; 北京大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本文提出StaMina状态维护方法，通过状态条件更新和可微循环学习视频计数中的状态转换，在SVCBench上显著提升计数精度，优于Counting-SFT。
AI 中文摘要
连续视频计数需要区分新观察与新增物体或已完成事件。我们提出StaMina（状态维护），通过状态条件更新来学习维护计数状态。循环视觉上下文支持识别；学习到的转换维护可见性、持久身份和已完成事件记录。可微循环在由计数端点约束的合法路径上训练事件转换；可见性和关联目标训练物体分支。多源流程将39.8K空间查询和补充事件标注组织成计数轨迹。在SVCBench上，我们评估了部分视频重叠和保留组链接标注的计数适应。在前缀重放（Full）和持续流式（Stream）下，4B和8B模型分别达到41.9/36.4和44.9/38.2的高斯精度准确率。8B模型在同一查询上比Counting-SFT提升10.9/3.2点。匹配图比较隔离了阶段条件和轨迹监督，评估训练目标与硬决策。在线视频基准和计数条件决策评估在线理解和任务资格。项目页面：此https URL
英文摘要
Continuous video counting requires distinguishing new observations from new objects or completed events. We introduce StaMina (State Maintenance), which learns to maintain counting state through state-conditioned updates. Recurrent visual context supports recognition; learned transitions maintain visibility, persistent identities, and completed-event records. A differentiable recurrence trains event transitions over legal paths constrained by count endpoints; visibility and association objectives train the object branch. A multi-source pipeline organizes 39.8K spatial queries and complementary event annotations into counting trajectories. On SVCBench, we evaluate counting adaptation with partial video overlap and held-out groups of linked annotations. Under prefix replay (Full) and persistent streaming (Stream), 4B and 8B models reach 41.9/36.4 and 44.9/38.2 Gaussian Precision Accuracy, respectively. The 8B model gains 10.9/3.2 points over Counting-SFT on the same queries. Matched-graph comparisons isolate phase conditioning and trajectory supervision, assessing training objectives alongside hard decisions. Online video benchmarks and count-conditioned decisions assess online understanding and task eligibility. Project Page: https://PLACEHOLDER.github.io/StaMina/
Comments28 pages, 7 figures. Project Page: https://PLACEHOLDER.github.io/StaMina/
