Towards Practical Real-Time Neural Video Compression (DCVC-RT)

CVPR 2025 paper on a neural video codec achieving 125fps at 1080p encode/decode with 21% bitrate savings versus H.266/VTM.

Link

• Visit resource

Related resources

• Puffer (Stanford research study)Overview of Stanford's Puffer, a live production research platform with real users studying ML-driven adaptive bitrate algorithms.
• YTLive: A Dataset of Real-World YouTube Live Streaming SessionsAcademic dataset paper capturing real-world YouTube live streaming session traces for research use.
• NUS VV-Streams QoE ModelSubjective QoE dataset and model for volumetric video geometry/texture compression from NUS research lab.
• RL-DASHReinforcement learning (Q-learning) based adaptive bitrate control simulated over ns3 and DASH.
• NetLLMSIGCOMM 2024 research adapting LLMs for networking tasks including adaptive bitrate streaming, viewport prediction, and job scheduling, wit…
• Testing DRM Playback Across Device FamiliesPractical guide and checklist for validating DRM playback matrices across different device families.
• FairPlay-iOS-DemoClient-side SPC/CKC FairPlay Streaming flow demo implementing Apple's FairPlay Streaming Programming Guide, MIT licensed.