Dual-Signal Adaptive KV-Cache Optimization for Long-Form Video Understanding in Vision-Language Models
This addresses the problem of inefficient memory usage in VLMs for long-form video understanding, enabling processing on consumer-grade hardware, though it is incremental as it builds on existing KV-cache optimization methods.
The paper tackled the memory bottleneck in Vision-Language Models (VLMs) when processing long-form video content by proposing Sali-Cache, a dual-signal adaptive KV-cache optimization framework, achieving a 2.20x compression ratio in memory usage while maintaining 100% accuracy on metrics like BLEU and ROUGE-L.
Vision-Language Models (VLMs) face a critical memory bottleneck when processing long-form video content due to the linear growth of the Key-Value (KV) cache with sequence length. Existing solutions predominantly employ reactive eviction strategies that compute full attention matrices before discarding tokens, resulting in substantial computational waste. We propose Sali-Cache, a novel a priori optimization framework that implements dual-signal adaptive caching through proactive memory management. By integrating a temporal filter based on optical flow analysis for detecting inter-frame redundancy and a spatial filter leveraging saliency detection for identifying visually significant regions, Sali-Cache intelligently manages memory allocation before entering computationally expensive attention operations. Experimental evaluation on the LLaVA 1.6 architecture demonstrates that our method achieves a 2.20x compression ratio in effective memory usage while maintaining 100% accuracy across BLEU, ROUGE-L, and Exact Match metrics. Furthermore, under identical memory budget constraints, Sali-Cache preserves context-rich features over extended temporal durations without degrading model performance, enabling efficient processing of long-form video content on consumer-grade hardware.