VideoMost Research, an R&D division based in Dubai, has recently been granted a US patent for its innovative video compression technology. This technology, which utilizes sketch concepts, neural networks, and machine learning, is set to revolutionize the way video content is transmitted and stored globally. With over 80% of global internet traffic being video, as reported by Cisco, the need for more efficient compression methods has never been more critical.
The patented method by VideoMost Research claims to triple the video compression factor compared to current standards such as H.265, VP9, and AV1, without compromising on quality. By converting the original video stream into a sketch-video that requires fewer resources, and then using a trained neural network to restore it to its original form upon decoding, the technology offers a promising solution to the challenges of video storage and transmission.
This machine learning-based codec not only reduces video bitrate and storage costs but also ensures high-quality streaming over narrowband mobile communication channels. Its robustness against connectivity disruptions and hardware-agnostic nature further enhance its applicability across various sectors. Potential applications include video portals, conferencing, streaming platforms, and more, highlighting the technology's versatility and potential impact on the digital content landscape.
With over 40 patents and 150 years of combined experience, the VideoMost Research team is at the forefront of video compression innovation. The company is currently seeking financial partnerships to further develop and distribute its ML codec, signaling a significant step forward in addressing the global demand for efficient video content management.
