AI RESEARCH
LiveVLM: Efficient Online Video Understanding via Streaming-Oriented KV Cache and Retrieval
arXiv CS.CV
•
ArXi:2505.15269v2 Announce Type: replace Recent developments in Video Large Language Models (Video LLMs) have enabled models to process hour-long videos and exhibit exceptional performance. Nonetheless, the Key-Value (KV) cache expands linearly over time, leading to substantial memory overhead and response delay--critical challenges in various real-world online applications, such as Deepseek services, autonomous driving and robotics. To mitigate these issues, we propose $\textbf{LiveVLM}$, a.