PCA: Persistence-Aware Compression and Aggregation for Fast Video Large Language Models

arXiv:2607.22726v1 Announce Type: new Abstract: Despite advances in Video Large Language Models (VLLMs) that have displayed promising outcomes in video understanding, the redundancy in the long-duration frames remains a hindrance to efficient reasoning. This paper introduces a training-free mathbf{P}ersistence-Aware mathbf{C}ompression and mathbf{A}ggregation (PCA) method designed to preserve high-fidelity raw visual information before the encoding stage. PCA can be built on arbitrary VLLMs and consists of two modules: 1) A Dynamic Downsampling (DD) module that adaptively removes redundant frames by analyzing frame-wise similarity. 2) A Persistence-Aware Motion Enhancement (PAME) module that enriches each selected keyframe by aggregating the temporal context of its neighbors, ensuring that essential information is preserved even after aggressive frame reduction. Our approach substantially reduces the computation of long-context modeling, while enhancing the performance of the baseline model. Extensive experiments demonstrate that PCA consistently outperforms existing state-of-the-art approaches in both efficiency and accuracy, achieving a speedup of 1.8imes to 2.5imes compared to the baseline VLLM. The code is open-sourced at https://github.com/Heisenberg10110/PCA.

Related

• CATS: Curvature Aware Temporal Selection for efficient long video understanding
• EchoPrune: Interpreting Redundancy as Temporal Echoes for Efficient VideoLLMs
• Gaussian Mixture Modeling for Event-Aware Visual Allocation in Long Video Understanding

Source: arXiv cs.CV | 2026-07-28