Beamr ML-Safe Lossless Compression Expands AV Data Capacity

Beamr ML-Safe Lossless Compression Expands AV Data Capacity

Beamr Imaging Ltd. has expanded its ML-Safe video data compression solution with lossless compression for autonomous vehicle applications. The company said the addition allows autonomous vehicle teams to reduce video data file sizes while preserving the information required by machine learning systems. The updated solution now supports both lossy compression, which is validated against the model’s own accuracy metrics, and lossless compression, which preserves data exactly at the bit level. For Beamr Imaging Ltd, the new capability addresses a specific challenge in managing large volumes of vehicle-generated video data without compromising the underlying information used by machine learning models.

Lossless Compression Preserves 12-Bit Bayer Data

The lossless capability is designed for autonomous vehicle teams that need to retain source data without introducing changes caused by compression. Beamr’s approach takes 12-bit Bayer data and compresses it with zero loss, meaning the resulting compressed representation can preserve the original information exactly. This differs from lossy compression, where the objective is to reduce file size while validating that model performance remains within acceptable accuracy metrics. By adding a bit-exact option, Beamr Imaging Ltd provides AV development teams with a compression method intended for workflows where preserving the captured data itself is a central requirement.

Higher Data-Capture Capacity Without New Hardware

The new lossless compression capability is also intended to address the infrastructure demands associated with collecting and retaining autonomous vehicle video. According to the announcement, compression of the 12-bit Bayer data can almost double data-capture capacity while preserving the captured information and avoiding the need for new hardware. That combination can allow AV teams to handle more data within existing capture infrastructure, while maintaining the source information needed for subsequent processing and machine learning workflows. The practical significance is that additional capacity can come from reducing the amount of storage required for each captured dataset rather than from expanding the underlying hardware footprint.

Industry Impact & Outlook

For the autonomous vehicle industry, the addition of bit-exact compression creates another option for managing increasingly demanding video-data workflows, particularly where original sensor information must be retained. The ability to increase capture capacity without adding hardware could affect how development teams approach storage and data collection when existing infrastructure is constrained. The distinction between model-validated lossy compression and zero-loss compression also gives AV teams a choice based on the requirements of a particular workflow. Beamr Imaging Ltd is scheduled to demonstrate the ML-Safe solution at AutoSens Europe in Barcelona from September 22–24, 2026, providing an opportunity to present the expanded capability directly to the automotive technology community.

Frequently Asked Questions

What did Beamr add to its ML-Safe video compression solution?
Beamr added lossless compression to its ML-Safe video data solution for autonomous vehicle applications, alongside its existing model-validated lossy compression capability. The lossless mode is designed to compress 12-bit Bayer data with zero loss, preserving the original information at the bit level. According to the announcement, the capability can almost double data-capture capacity while maintaining the captured data and requiring no new hardware. Beamr is scheduled to demonstrate the expanded solution at AutoSens Europe in Barcelona from September 22–24, 2026.