Beamr (Nasdaq: BMR), a leader in content-adaptive video compression, an NVIDIA and AWS partener trusted by top technology companies - is working with tier-1 suppliers and OEMs in the AV and ADAS industry to deliver ML-safe video data compression at petabyte scale.
Beamr delivers up to 50% video data file size reduction, which immediately translates into reduced storage footprint, I/O time, transfer costs, and downstream data loading. ML-safe compression was benchmarked across the AV pipeline, with less than 2% mean average precision difference; localization, and confidence were well preserved. Testing with world foundation model captioning pipelines confirmed no measurable impact on pipeline outputs beyond inherent model variability.
The patented compression technology, Content-Adaptive Bitrate (CABR), integrates directly into existing workflows for both real-world and synthetic video data, and for already-encoded and newly ingested footage. Deployment is flexible (public or private cloud, FFmpeg plug-in or SDK). CABR runs on NVIDIA GPUs already in your stack. You can ingest any input, output is in standard codecs: AVC, HEVC, and AV1.
