A Simple Baseline for Streaming Video Understanding

TL;DR

A simple sliding-window baseline using recent frames with off-the-shelf VLMs matches or exceeds complex streaming video models, challenging the need for intricate memory mechanisms.

Contribution

Demonstrates that a straightforward recent-frame approach can outperform complex memory-based streaming video models, urging a reevaluation of current benchmarking practices.

Findings

SimpleStream achieves 67.7% accuracy on OVO-Bench with only 4 frames.

Longer context benefits depend on the backbone model, not always improving performance.

Adding historical context can improve recall but may weaken real-time perception.

Abstract

Recent streaming video understanding methods increasingly rely on complex memory mechanisms to handle long video streams. We challenge this trend with a simple finding: a sliding-window baseline that feeds only the most recent N frames to an off-the-shelf VLM already matches or surpasses published streaming models. We formalize this baseline as SimpleStream and evaluate it against 13 major offline and online video LLM baselines on OVO-Bench and StreamingBench. Despite its simplicity, SimpleStream delivers consistently strong performance. With only 4 recent frames, it reaches 67.7% average accuracy on OVO-Bench and 80.59% on StreamingBench. Controlled ablations further show that the value of longer context is backbone-dependent rather than uniformly increasing with model scale, and reveal a consistent perception-memory trade-off: adding more historical context can improve recall, but often weakens real-time perception. This suggests that stronger memory, retrieval, or compression modules should not be taken as evidence of progress unless they clearly outperform SimpleStream under the same protocol. We therefore argue that future streaming benchmarks should separate recent-scene perception from long-range memory, so that performance improvements from added complexity can be evaluated more clearly.