@inproceedings{657e327611f2407195fbccbce2ecfdf7,
title = "S3-Prune: Stability-Aware Token Budgeting for Long-Form Video-Language Models",
abstract = "The deployment of long-form video-language models (Video-VLMs) is fundamentally constrained by the prohibitive scaling of attention complexity and KV-cache overhead as temporal and spatial resolutions increase. Existing pruning methods rely on local frame-level heuristics, which fail to capture shifting information density and often result in cumulative budget misallocation over long horizons. To address these challenges, we propose S3-prune, a stability-aware token budgeting framework that jointly models spatio-temporal demand while ensuring consistent allocation over time. S3-prune characterizes token demand by integrating spatial uncertainty (SU) from patch-level embedding shifts and segment transition (ST) from inter-segment semantic variations. To mitigate temporal instability, we introduce stability accumulation (SA) via a Kalman-filtered latent estimator, which smoothens stochastic spikes and provides uncertainty-aware margins for adaptive budgeting. Guided by these stabilized budgets, S3-prune executes a hierarchical two-stage selection process. Extensive evaluations across seven benchmarks and various backbones demonstrate that S3-prune significantly advances the efficiency-accuracy frontier, reducing inference latency while remaining robust even under aggressive token retention settings.",
keywords = "Efficient Inference, Temporal Stabilization, Token Pruning, Video Language Model",
author = "Gilha Lee and Seungil Lee and Hyun Kim",
note = "Publisher Copyright: {\textcopyright} The Author(s), under exclusive license to Springer Nature Switzerland AG 2026.; 19th European Conference on Computer Vision, ECCV 2026 ; Conference date: 08-09-2026 Through 12-09-2026",
year = "2026",
doi = "10.1007/978-3-032-37252-9\_27",
language = "English",
isbn = "9783032372512",
series = "Lecture Notes in Computer Science",
publisher = "Springer Science and Business Media Deutschland GmbH",
pages = "484--501",
editor = "Paolo Favaro and Zuzana Kukelova and Atsuto Maki and Anna Rohrbach and Konrad Schindler and Federico Tombari",
booktitle = "Computer Vision {\textendash} ECCV 2026 - 19th European Conference, 2026, Proceedings",
}