Research map / Vision-language and multimodal / Multimodal LLMs
OmniStream: Mastering Perception, Reconstruction and Action in Continuous Streams
Yibin Yan, Jilan Xu, Shangzhe Di, Haoning Wu, Weidi Xie
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 |
|---|---|---|---|
| add video data to SSL (ablation) | OmniStream · SSv2 | accuracy: 63.0 -> 69.3 | Improved |
| add video data to SSL (ablation) | OmniStream · DAVIS'17 | J&F: 67.7 -> 71.6 | Improved |
| add video data to SSL (ablation) | OmniStream · CALVIN | avg. length: 3.42 -> 3.80 | Improved |
| add video data to SSL (ablation) | OmniStream · VSI-Bench | accuracy: 57.9 -> 57.3 | Hurt |
| add streaming geometric reconstruction (ablation) | OmniStream · NYUv2 | RMSE: 0.471 -> 0.379 | Improved |
| add streaming geometric reconstruction (ablation) | OmniStream · ADE20K | mIoU: 42.3 -> 49.6 | Improved |
| add streaming geometric reconstruction (ablation) | OmniStream · VSI-Bench | accuracy: 52.5 -> 57.3 | Improved |
| add streaming geometric reconstruction (ablation) | OmniStream · CALVIN | avg. length: 3.34 -> 3.80 | Improved |
| add captioning vision-language alignment (ablation) | OmniStream · VideoMME | accuracy: 45.0 -> 54.1 | Improved |
| add captioning vision-language alignment (ablation) | OmniStream · VSI-Bench | accuracy: 44.9 -> 57.3 | Improved |
| add captioning vision-language alignment (ablation) | OmniStream · CALVIN | avg. length: 2.38 -> 3.80 | Improved |
| use causal KV-cache instead of bidirectional recomputation (ablation) | OmniStream · per-frame latency at T=64 (s) | latency: 0.998 -> 0.067 | Improved |
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.
- SpaceMind — SpaceMind: Camera-Guided Modality Fusion for Spatial Reasoning in Vision-Language Models (CVPR 2026) · better on 1 of 1 shared result
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
- IGGT4D: Streaming 4D Instance-Grounded Geometry Transformer (NeurIPS 2026 · similarity 0.86)
- StreamOV: Streaming Omni-Video Understanding via Evidence-Guided Memory and Response Triggering (NeurIPS 2026 · similarity 0.86)
- ConsisVLA-4D: Advancing Spatiotemporal Consistency in Efficient 3D-Perception and 4D-Reasoning for Robotic Manipulation (CVPR 2026 · similarity 0.86)
- StreamForest: Efficient Online Video Understanding with Persistent Event Memory (NeurIPS 2025 · similarity 0.86)
- MVVBench: Benchmarking 4D Reasoning in Vision-Language Models (NeurIPS 2026 · similarity 0.85)
- FrameVGGT: Coherence-Preserving Memory for Bounded Streaming Geometry (NeurIPS 2026 · similarity 0.85)
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.
