AI Paper Atlas

Research map / Video / Video-language understanding

RIVER: Real-time Video Interaction Benchmark

Yansong Shi, Qingsong Zhao, Tianxiang Jiang, Xiangyu Zeng, Yi Wang, Limin Wang

ICLR 2026 · Video › Video-language understanding · approach: Event and anomaly reasoning in video

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 changedWhereMeasuredEffect
add RIVER pro-response fine-tuning data to VideoLLM-Online (2fps)VideoLLM-Online · RIVER Pro-Response InstantMC accuracy: 23.88 -> 33.28Improved
raise frame rate from 2fps to 4fps for VideoLLM-Online+RIVERVideoLLM-Online+RIVER · RIVER Pro-Response InstantMC accuracy: 33.28 -> 35.16Improved
add long-short term memory and sliding window to VideoChat-FlashVideoChat-Flash · RIVER Retro-MemoryMC accuracy: 41.51 -> 45.75Improved
add long-short term memory and sliding window to InternVL2.5InternVL2.5 · RIVER Live-PerceptionMC accuracy: 43.65 -> 58.84Improved
add long-short term memory and sliding window to InternVL2.5InternVL2.5 · RIVER Retro-MemoryMC accuracy: 42.68 -> 39.72Hurt
add long-short term memory and sliding window to Llava-VideoLlava-Video · RIVER Retro-MemoryMC accuracy: 46.00 -> 42.71Hurt
add long-short term memory and sliding window to Llava-VideoLlava-Video · RIVER Live-PerceptionMC accuracy: 41.00 -> 51.38Improved
add long-short term memory and sliding window to VideoChat2VideoChat2 · RIVER Live-PerceptionMC accuracy: 29.28 -> 41.16Improved
add memory module to MLLM for long recall questions (ablation)MLLM (Figure 5) · RIVER Retro-Memorydecay slope: 12% lower decay slopeImproved
increase recall interval from short to very longGPT-4o · RIVER Retro-MemoryMC accuracy: 63.26 -> 52.21Hurt
increase recall interval from short to very longLlava-Video (16 frames) · RIVER Retro-MemoryMC accuracy: 49.17 -> 42.54Hurt
add online pipeline to VideoChat-Flash on background cuesVideoChat-Flash · RIVER cue typesMC accuracy (BC): 44.92 -> 54.10Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most 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

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy