Beamr Imaging, a leader in video optimization technology and solutions, announced it will launch a high-performance, high-quality video compression solution designed for autonomous vehicles at NVIDIA GTC Paris, taking place June 10-12, 2025, as part of Viva Technology 2025, Europe’s biggest startup and tech event.
In the development of autonomous driving, video is the dominant data type. A single vehicle produces terabytes of video data daily, and training a single autonomous model may require tens to hundreds of petabytes. Beamr’s proprietary technology, with a proven track record in high-efficiency video compression trusted by global media players, now addresses a pressing, costly challenge for autonomous vehicles and machine learning teams: managing video data at scale, including long-term storage and the significant infrastructure investment required.
Beamr’s Content-Adaptive Bitrate (CABR) technology, built on NVIDIA accelerated computing, reduces real-world autonomous driving and synthetic video file sizes by up to 50%. This is achieved while preserving visual quality and critical visual features essential for training autonomous driving models. By addressing key storage, compute, and bandwidth constraints in AI pipelines, it significantly reduces operational costs.
Also Read: Sony Semiconductor Solutions to Release Stacked SPAD Depth Sensor for Automotive LiDAR Applications
Video capturing for autonomous driving models starts with Advanced Driver Assistance Systems (ADAS), recording real-world driving footage ingested into data centers, where video volumes scale rapidly. Yet, real-world data alone is insufficient for training models that must perform reliably across a wide spectrum of scenarios, including rare edge cases. To address this, a vast amount of synthetic video is generated by platforms such as NVIDIA Omniverse and NVIDIA Cosmos™ world foundation models, helping to amplify training data.
“Autonomous vehicle companies are under mounting pressure from rising video storage demands and infrastructure costs,” said Sharon Carmel, Beamr CEO. “Our content-adaptive technology, accelerated by GPUs, delivers highly efficient compression while maintaining visual quality across a variety of scenarios – both for human perception and machine vision, and in both real-world and synthetic video.”
SOURCE: GlobeNewswire