Beamr, a leader in video optimization technology and solutions announced that it released a White Paper detailing how its Content Adaptive Bitrate (CABR) optimized encoding solution, which reduces video size but not perceptual quality, can make video used for vision AI easier to handle, thus reducing workflow complexity. Beamr is advancing on a new front – and reveals its capability to support Machine Learning for video.
Recommended AI News: Riding on the Generative AI Hype, CDP Needs a New Definition in 2024
Machine Learning (ML) for video processing is a field expanding at a fast pace, with a market already estimated at billions of dollars. One of the biggest pain points for ML players is managing extremely large video files and libraries. As the cluster of files grows bigger, they are faced with the challenging task of storing and transferring them at an increasing cost.
Results of a new case study show it may be possible to address the challenge and cut processing costs using Beamr’s technology.
Screen shot form Beamr Machine Learning experiment showing that true detection results are unaffected by replacing the source file (left) with the smaller, easier-to-transfer, optimized file (right).
In the new White Paper, Beamr shows that video files that were slimmed down by 40% on average – without losing their perceptual quality due to Beamr’s CABR technology – keep ML results unaffected.
Recommended: Predictions Series 2022: AiThority Interview with Peter Stone, Executive Director at Sony AI
The tests were conducted on NVIDIA DeepStream SDK – a tool for AI-based multi-sensor processing, video, audio and image understanding, which was a natural choice for Beamr as an NVIDIA Metropolis partner.
Tamar Shoham Beamr CTO, said: “We decided to try out the Nvidia DeepStream SDK, which enables vision AI applications and services, combined with the recent Nvidia encoder enhanced with Beamr CABR. In the experiment, we demonstrated that the results of the DeepStream were unaffected by the video optimization process”.
“We are thankful to the Nvidia DeepStream team for supporting our research”, Shoham Added.
The results presented in the White Paper show that Beamr’s patent-proven and award winning technology – Content Adaptive Bitrate – can be applied to videos that undergo ML tasks such as object detection. In future work, Beamr plans to investigate the further potential benefits obtained when CABR is incorporated at the training stage and expand the experiments to include more model types and ML tasks.
Recommended: Predictions Series 2022: AiThority Interview with David Low, CMO at Talkwalker
[To share your insights with us, please write to sghosh@martechseries.com]
Comments are closed.