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

arXiv: https://arxiv.org/abs/2607.22726 · HTML: https://mlsanity.com/paper/2607.22726/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 23 of its 38 references that are in the map.
- First posted on arXiv 22 Jul 2026.

## Builds on (most cited first, 23 of 23)

- [LLaMA: Open and Efficient Foundation Language Models](https://mlsanity.com/paper/2302.13971/) (2023)
- [Qwen2-VL: Enhancing Vision-Language Model's Perception of the World at Any Resolution](https://mlsanity.com/paper/2409.12191/) (2024)
- [LLaVA-OneVision: Easy Visual Task Transfer](https://mlsanity.com/paper/2408.03326/) (2024)
- [Expanding Performance Boundaries of Open-Source Multimodal Models with Model, Data, and Test-Time Scaling](https://mlsanity.com/paper/2412.05271/) (2024)
- [Video-LLaMA: An Instruction-tuned Audio-Visual Language Model for Video Understanding](https://mlsanity.com/paper/2306.02858/) (2023)
- [Video-MME: The First-Ever Comprehensive Evaluation Benchmark of Multi-modal LLMs in Video Analysis](https://mlsanity.com/paper/2405.21075/) (2024)
- [MVBench: A Comprehensive Multi-modal Video Understanding Benchmark](https://mlsanity.com/paper/2311.17005/) (2023)
- [Token Merging: Your ViT But Faster](https://mlsanity.com/paper/2210.09461/) (2022)
- [VideoLLaMA 2: Advancing Spatial-Temporal Modeling and Audio Understanding in Video-LLMs](https://mlsanity.com/paper/2406.07476/) (2024)
- [An Image is Worth 1/2 Tokens After Layer 2: Plug-and-Play Inference Acceleration for Large Vision-Language Models](https://mlsanity.com/paper/2403.06764/) (2024)
- [Towards Automatic Learning of Procedures from Web Instructional Videos](https://mlsanity.com/paper/1703.09788/) (2017)
- [InternVideo: General Video Foundation Models via Generative and Discriminative Learning](https://mlsanity.com/paper/2212.03191/) (2022)
- [LLaVA-PruMerge: Adaptive Token Reduction for Efficient Large Multimodal Models](https://mlsanity.com/paper/2403.15388/) (2024)
- [VideoAgent: Long-form Video Understanding with Large Language Model as Agent](https://mlsanity.com/paper/2403.10517/) (2024)
- [Perception Test: A Diagnostic Benchmark for Multimodal Video Models](https://mlsanity.com/paper/2305.13786/) (2023)
- [Apollo: An Exploration of Video Understanding in Large Multimodal Models](https://mlsanity.com/paper/2412.10360/) (2024)
- [LION : Empowering Multimodal Large Language Model with Dual-Level Visual Knowledge](https://mlsanity.com/paper/2311.11860/) (2023)
- [TESTA: Temporal-Spatial Token Aggregation for Long-form Video-Language Understanding](https://mlsanity.com/paper/2310.19060/) (2023)
- [Accelerating Multimodal Large Language Models by Searching Optimal Vision Token Reduction](https://mlsanity.com/paper/2412.00556/) (2024)
- [Human Motion Video Generation: A Survey](https://mlsanity.com/paper/2509.03883/) (2025)
- [PhysLLM: Harnessing Large Language Models for Cross-Modal Remote Physiological Sensing](https://mlsanity.com/paper/2505.03621/) (2025)
- [PHASE-Net: Physics-Grounded Harmonic Attention System for Efficient Remote Photoplethysmography Measurement](https://mlsanity.com/paper/2509.24850/) (2025)
- [OmniOPSD: Rationale-Privileged On-Policy Self-Distillation for Affective Computing](https://mlsanity.com/paper/2606.15920/) (2026)

## Abstract (arXiv)

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 Persistence-Aware Compression and Aggregation (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 1.8× to 2.5× compared to the baseline VLLM. The code is open-sourced at https://github.com/Heisenberg10110/PCA.