TurnWise improves streaming video RL with turn-level credit
TurnWise assigns turn-level advantages using silence-completed prefix scoring to reduce redundancy in streaming video policy learning
Stories per day · 30 days
2 RSS
Relatedbenchmarks2video-understanding2agents1deltastreamer1grpo1hallucination1memory1post-training1replay1rl1spatial-vlm1supervision1temporal-supervision1turn-wise1
9 stories, newest first · one card per story
Stories tagged #streaming-video
Visible-State Supervision targets under-specified streaming video recognition
Under a matched recent-frame setup, ViSS improves OVO-Bench and helps hallucination detection most, while staying comparable on StreamingBench.
ChronoReplay reduces interface lock-in in spatial VLMs
One checkpoint improves OVO-S-Bench, OST-Bench and UCS-Bench together, including a 13.41-point gain on OST-Bench.
DeltaStreamer: Towards Human-Like Agentic Streaming Video Understanding
DeltaStreamer combines change-aware memory and targeted replay for streaming video question answering.
OneStreamer: Unifying Perception, Memory, and Proactive Response in Streaming Video Interaction
OneStreamer-4B unifies streaming video perception, caption memory, and proactive responses, topping eight benchmarks.
APM-Bench: Benchmarking Cross-session Persistent Memory for Egocentric Streaming Video Assistants
APM-Bench evaluates persistent memory for streaming video assistants across 104 multi-session trajectories.
TRACE: Temporal Audit and Condition-aware Evaluation of Streaming Video Understanding
TRACE benchmarks streaming video models on timing, triggers, workload, and reliability, not accuracy alone.
ReactVAU: A Slow-Fast Decoupled Framework for Streaming Video Anomaly Understanding
ReactVAU is a slow-fast streaming framework for real-time video anomaly understanding that reserves heavyweight MLLM reasoning for suspicious events, improving efficiency.
ShallowStream: Index Shallow then Answer Deep for Streaming Video Understanding
ShallowStream builds streaming-video retrieval indexes from shallow MLLM layers, cutting per-frame prefill latency by up to 52.1x.