Research map / Video / Video-language understanding
STRIDE: When to Speak Meets Sequence Denoising for Streaming Video Understanding
Junho Kim, Hosu Lee, James Rehg, Minsu Kim, Yong Man Ro
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| replace autoregressive binary activation with masked diffusion (STRIDE) | Qwen3-VL-2B activation model · ET-Bench | avg F1: 24.3 -> 32.6 | Improved |
| STRIDE vs Baseline-AR on Qwen3-VL-8B | Qwen3-VL-8B · OVO-Bench | overall avg: 52.81 -> 59.07 | Improved |
| STRIDE vs Baseline-AR on Qwen3-VL-8B | Qwen3-VL-8B · StreamingBench | overall avg: 57.12 -> 59.29 | Improved |
| STRIDE vs Baseline-AR on forward active responding | Qwen3-VL-8B · OVO-Bench | forward active responding avg: 42.73 -> 59.70 | Improved |
| add STRIDE to Gemma3-4B | Gemma3-4B · OVO-Bench | overall avg: 42.13 -> 50.51 | Improved |
| add STRIDE to InternVL3-8B | InternVL3-8B · StreamingBench | overall avg: 53.97 -> 57.58 | Improved |
| use structured span masking instead of independent masking (ablation) | STRIDE · ET-Bench | avg F1: 7.2 -> 21.1 | Improved |
| add full masking and span unmasking to span masking (ablation) | STRIDE · ET-Bench | avg F1: 21.1 -> 32.6 | Improved |
| add sequence duplication (ablation) | STRIDE · ET-Bench | avg F1: 22.9 -> 32.6 | Improved |
| use selective re-masking instead of last-only (ablation) | STRIDE · ET-Bench | avg F1: 22.6 -> 32.6 | Improved |
| set retention threshold tau to 0 (retain all) (ablation) | STRIDE · ET-Bench TVG | F1: -19.7 pt | Hurt |
| increase denoising steps K beyond 8 (ablation) | STRIDE · ET-Bench | mean F1 | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- Dispider — Dispider: Enabling Video LLMs with Active Real-Time Interaction via Disentangled Perception, Decision, and Reaction (CVPR 2025) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- LiveStar: Live Streaming Assistant for Real-World Online Video Understanding (NeurIPS 2025 · similarity 0.88)
- StreamBridge: Turning Your Offline Video Large Language Model into a Proactive Streaming Assistant (NeurIPS 2025 · similarity 0.88)
- Before It Fades: Reinforcing Temporal Representations at Inference Time in VideoLLMs (NeurIPS 2026 · similarity 0.87)
- What Should a Streaming Video Model Remember? (NeurIPS 2026 · similarity 0.86)
- SVBench: A Benchmark with Temporal Multi-Turn Dialogues for Streaming Video Understanding (ICLR 2025 · similarity 0.86)
- StreamGaze: Gaze-Guided Temporal Reasoning and Proactive Understanding in Streaming Videos (NeurIPS 2026 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
