---
title: "Beamr Edges Ahead on Video Stack Word"
type: "News"
locale: "en"
url: "https://longbridge.com/en/news/287087986.md"
description: "Beamr Imaging Ltd. (NASDAQ: BMR) announced its ML-safe video data stack for autonomous vehicles at AutoSens USA 2026, aiming for a 50% file size reduction without performance loss. The technology supports scalable training pipelines and has shown a 30.7% reduction in depth error for AI models using Beamr-compressed footage. BMR shares rose 1% to $3.19."
datetime: "2026-05-20T15:02:21.000Z"
locales:
  - [zh-CN](https://longbridge.com/zh-CN/news/287087986.md)
  - [en](https://longbridge.com/en/news/287087986.md)
  - [zh-HK](https://longbridge.com/zh-HK/news/287087986.md)
---

# Beamr Edges Ahead on Video Stack Word

Beamr Imaging Ltd. (NASDAQ: BMR) shares started out Wednesday slightly in the green. The company, claiming to be a leader in video optimization technology and solutions, today announced that it will present its ML-safe video data stack for autonomous vehicles (AV) at AutoSens USA 2026, held from June 9–11 in Detroit. AutoSens is an AV and ADAS engineering-focused conference, attracting professionals and leaders from OEMs, Tier 1, and supply chain companies.

As autonomous driving advances and integrates with Physical AI applications, the training and validation pipelines performance determine how far the model development can advance. Perception, validation, and simulation engineers need a proven compression approach that will support these scalable pipelines, often reaching hundreds of petabytes.

Beamr is offering its ML-safe video data stack, enabling up to 50% file size reduction without degrading model performance. The stack was validated from perception models through to world foundation models, on real-world and synthetic footage.

Recently, Beamr demonstrated that video compressed with its patented content-adaptive bitrate (CABR) technology can be used as an augmentation step in AI model training. The study demonstrated that treating compression as a training strategy allows it to scale efficiently while preserving the perception accuracy ML systems depend on. A state-of-the-art depth model trained with Beamr-compressed footage, delivered a 30.7% reduction in depth error on vulnerable road users (VRUs), including pedestrians and motorcyclists, while compressing the training data by 35.2% compared to baseline.

BMR shares began Wednesday up three cents, or 1%, to $3.19.

### Related Stocks

- [BMR.US](https://longbridge.com/en/quote/BMR.US.md)

## Related News & Research

- [Beamr to Present its ML-Safe Video Data Stack to AV Engineers at AutoSens USA | BMR Stock News](https://longbridge.com/en/news/287057657.md)
- [Beamr Validates ML-Safe Compression for dSPACE Data Logging | BMR Stock News](https://longbridge.com/en/news/283382024.md)
- [Shamrock Capital Launches Fourth Content Strategy Fund Since 2015 With $813 Million in Commitments](https://longbridge.com/en/news/287065503.md)
- [AI face is taking over — and driving plastic surgeons crazy](https://longbridge.com/en/news/286641783.md)
- [Uber-Waymo Partnership On The Rocks? Exec Shares 'Scary' SF Robotaxi Moment, Slam AV Rollout](https://longbridge.com/en/news/286542868.md)