OASIS: On-Demand Hierarchical Event Memory for Streaming Video Reasoning

TL;DR

OASIS introduces a hierarchical, on-demand memory retrieval framework for streaming video reasoning, improving accuracy and efficiency by focusing on relevant evidence through semantic, intent-driven retrieval.

Contribution

It presents a novel, training-free, plug-and-play hierarchical retrieval mechanism that enhances long-horizon reasoning in streaming video models.

Findings

Achieves strong gains in long-horizon accuracy.

Reduces token cost and request delay.

Improves compositional reasoning across benchmarks.

Abstract

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.