StreamPro: From Reactive Perception to Proactive Decision-Making in Streaming Video

TL;DR

StreamPro introduces a new benchmark and training framework for proactive streaming video understanding, enabling models to make early, reliable decisions under partial observations.

Contribution

The paper presents StreamPro-Bench for evaluating proactive reasoning and proposes StreamPro, a two-stage training method with novel loss and reward designs.

Findings

StreamPro significantly outperforms previous methods on the new benchmark.

It achieves 41.5 in proactive performance, compared to 10.4 of the previous best.

On StreamingBench-RTVU, it reaches 78.9, demonstrating strong real-time streaming capabilities.

Abstract

Proactive streaming video understanding requires models to continuously process video streams and decide when to respond, rather than merely what to respond. This naturally introduces a decision-making problem under partial observations, where models must balance early prediction against sufficient evidence. However, existing benchmarks largely follow a "see-then-answer" paradigm, where responses are triggered only after explicit evidence appears, effectively reducing proactive reasoning to delayed perception. As a result, they fail to evaluate a model's ability to make timely and reliable decisions under incomplete observations. Moreover, training proactive models is inherently challenging due to the extreme imbalance between silence and response signals in streaming trajectories, as well as the need to jointly optimize response correctness and timing. To address these challenges, we introduce StreamPro-Bench, a new benchmark that evaluates streaming models from three complementary perspectives: Perception Understanding, Temporal Reasoning, and Proactive Agency, where the last measures a model's ability to make early yet reliable decisions under partial observations. We further propose StreamPro, a two-stage training framework for proactive learning. First, we introduce CB-Stream Loss to mitigate the severe supervision imbalance during supervised fine-tuning (SFT). Then, we apply Group Relative Policy Optimization (GRPO) with a multi-grained reward design that involves both turn-level and trajectory-level rewards. Experiments show that StreamPro significantly improves proactive performance. On StreamPro-Bench, it achieves 41.5, substantially outperforming the previous best (10.4), while also maintaining strong performance on real-time streaming benchmarks, achieving 78.9 on StreamingBench-RTVU.