OASIS: On-Demand Hierarchical Event Memory for Streaming Video Reasoning
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866910143611928576 |
|---|---|
| author | Liang, Zhijia Li, Jiaming Chen, Weikai Zhang, Yanhao Lu, Haonan Li, Guanbin |
| author_facet | Liang, Zhijia Li, Jiaming Chen, Weikai Zhang, Yanhao Lu, Haonan Li, Guanbin |
| contents | Streaming video reasoning requires models to operate in a setting where history grows without bound while meaningful evidence remains scarce. In such a landscape, relevant signal is like an oasis-small, critical, and easily lost in a desert of redundancy. Enlarging memory only widens the desert; aggressive compression dries up the oasis. The real difficulty lies in discovering where to look, not how much to remember. We therefore introduce OASIS, a novel framework for streaming video reasoning that tackles this challenge through structured, on-demand retrieval. It organizes streaming history into hierarchical events and performs reasoning as controlled refinement-short-context inference first, followed by semantically grounded retrieval only when uncertainty arises. As the retrieval is driven by high-level intent rather than embedding similarity, the retrieved memory is substantially more accurate and less noisy. Additionally, the mechanism is plug-and-play, training-free, and readily attaches to different streaming MLLM backbones. Experiments across multiple benchmarks and backbones show that OASIS achieves strong gains in long-horizon accuracy and compositional reasoning with bounded token cost and low request delay. Code is available at https://github.com/Solus-sano/OASIS. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2604_17052 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | OASIS: On-Demand Hierarchical Event Memory for Streaming Video Reasoning Liang, Zhijia Li, Jiaming Chen, Weikai Zhang, Yanhao Lu, Haonan Li, Guanbin Computer Vision and Pattern Recognition Streaming video reasoning requires models to operate in a setting where history grows without bound while meaningful evidence remains scarce. In such a landscape, relevant signal is like an oasis-small, critical, and easily lost in a desert of redundancy. Enlarging memory only widens the desert; aggressive compression dries up the oasis. The real difficulty lies in discovering where to look, not how much to remember. We therefore introduce OASIS, a novel framework for streaming video reasoning that tackles this challenge through structured, on-demand retrieval. It organizes streaming history into hierarchical events and performs reasoning as controlled refinement-short-context inference first, followed by semantically grounded retrieval only when uncertainty arises. As the retrieval is driven by high-level intent rather than embedding similarity, the retrieved memory is substantially more accurate and less noisy. Additionally, the mechanism is plug-and-play, training-free, and readily attaches to different streaming MLLM backbones. Experiments across multiple benchmarks and backbones show that OASIS achieves strong gains in long-horizon accuracy and compositional reasoning with bounded token cost and low request delay. Code is available at https://github.com/Solus-sano/OASIS. |
| title | OASIS: On-Demand Hierarchical Event Memory for Streaming Video Reasoning |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2604.17052 |