Beamr to Present ML-Safe Video Stack at AutoSens USA with 50% Compression
BMR•Beamr will showcase its ML-safe video data stack for autonomous vehicles at AutoSens USA 2026, delivering up to 50% file size reduction without degrading model performance. A depth model trained on Beamr-compressed data cut depth error on vulnerable road users by 30.7% while reducing training data size by 35.2%.
1. AutoSens USA Presentation
Beamr Imaging will present its ML-safe video data stack for autonomous vehicles at AutoSens USA 2026, held June 9–11 in Detroit. Attendees can meet Beamr’s team at booth 315 to explore integration of the compression solution into AV development pipelines.
2. ML-Safe Compression Capabilities
Beamr’s content-adaptive bitrate (CABR) technology achieves up to 50% reduction in video file size without degrading model performance. The stack has been validated from perception models through world foundation models on both real-world and synthetic footage.
3. AI Model Training Benefits
A state-of-the-art depth model trained using Beamr-compressed footage delivered a 30.7% reduction in depth error on vulnerable road users while cutting training data volume by 35.2%, demonstrating compression as an effective AI training augmentation strategy.
