Abstract
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/
Keywords
Subject
Publication details
- Journal
- Not available
- Open access
- Green open access
Cite this article
APA 7
Liu, P., Lyu, D., Niu, J., Shi, Z., Xie, J., & Liu, S. (2026). When Should the Count Change? Learning State Maintenance for Causal Video Counting. https://omanscience.com/en/articles/when-should-the-count-change-learning-state-maintenance-for-causal-video-counting
MLA 9
Liu, Pengyiang, et al. "When Should the Count Change? Learning State Maintenance for Causal Video Counting." https://omanscience.com/en/articles/when-should-the-count-change-learning-state-maintenance-for-causal-video-counting.
Chicago (author–date)
Liu, Pengyiang, Dongyue Lyu, Junbo Niu, Zhongyue Shi, Jiahao Xie, and Si Liu. 2026. "When Should the Count Change? Learning State Maintenance for Causal Video Counting." https://omanscience.com/en/articles/when-should-the-count-change-learning-state-maintenance-for-causal-video-counting.
Harvard
Liu, P., Lyu, D., Niu, J., Shi, Z., Xie, J. and Liu, S. (2026) 'When Should the Count Change? Learning State Maintenance for Causal Video Counting', Available at: https://omanscience.com/en/articles/when-should-the-count-change-learning-state-maintenance-for-causal-video-counting.
Vancouver
Liu P, Lyu D, Niu J, Shi Z, Xie J, Liu S. When Should the Count Change? Learning State Maintenance for Causal Video Counting. https://omanscience.com/en/articles/when-should-the-count-change-learning-state-maintenance-for-causal-video-counting
IEEE
P. Liu, D. Lyu, J. Niu, Z. Shi, J. Xie, and S. Liu, "When Should the Count Change? Learning State Maintenance for Causal Video Counting," https://omanscience.com/en/articles/when-should-the-count-change-learning-state-maintenance-for-causal-video-counting.