Deep learning for video provider, iSIZE, announced in a company press release that it has entered a partnership with Intel with the aim to lower costs, speed up and deliver video content creation, while also reducing energy consumption.
Sergio Grce, CEO of iSIZE wrote how the company is proud to showcase its work with Intel by saying: “We are proud to showcase our work with Intel to further optimise our AI precoding solution by leveraging the vast capabilities of Intel Xeon Scalable processors at the edge and cloud. iSIZE is able to bring significant bandwidth savings and reduced energy needs to a variety of workloads, such as VoD, live streaming, gaming, and IoT. We are looking forward to continuing our strategic partnership with Intel and seeing our deep perceptual optimisation grow to its fullest potential for customers.”
In the modern digital age, businesses and consumers are increasingly looking to get items and services quicker while at a better cost. COVID-19 has put pressure on businesses and reduced social interactions as people have been urged to stay home and also work from home. As a result, the demand for tools and services such as video streaming and content creation to interact with people virtually has soared.
iSIZE proprietary deep perceptual optimizer uses deep learning to train its AI model to ‘see with the human eye’ in order to optimise video quality with the goal to offer significant bitrate savings for all video encoding standards, including AVC, HEVC and AV1.
The deep learning for video providers technology is deployed as an add-on feature to conventional video encoding pipelines, without requiring any changes in the streaming process or the client devices, according to iSIZE. This can result in substantial bandwidth, energy, and cost savings for VoD and live streaming services, broadcasters, and the end consumers.
This could become an increasingly sought after technology as we switch to a digital economy with more people taking meetings virtually and more video content appearing and being developed every day.
It could become in particular use for VR and AR with companies such as Apple taking bets that it could potentially be the capability consumers are going to use by purchasing NextVR back in May 2020.
The deep learning for video provider claims its BitSave precoding technology can achieve up to 5x increased speed performance for up to 25% bitrate savings on Intel processors.
iSIZE claimed that this translates to a significant total cost of ownership savings for video distribution at similar video quality while giving the company a strong economic argument for commercial deployment of enhanced end-user experiences.
The deep learning for video provider wrote that this has been accomplished by working closely with Intel to optimise its AI models for Intel Xeon Scalable processors with Deep Learning Boost (Intel DL Boost) and the Intel Distribution of OpenVINO toolkit which is powered by oneAPI.
Nagesh Puppala, Senior Director for the Visual Infrastructure Division at Intel Data Platforms Group commented on what the work its carried out with iSIZE has allowed the company to accomplish by saying: “Video creation and consumption is growing exponentially. Our work with iSIZE on software optimizations using Intel® DL Boost shows how companies can take advantage of integrated AI advancements in Intel Xeon Scalable processors to reduce the compute power needed to run AI.”
The deep learning video provider explained how video streamings total cost of ownership is now dominated by the distribution costs for premium high-resolution content.
iSIZE claims that with this partnership with Intel, its customers, including Communication Service Providers, can now realise even better cost savings using its iSIZE BitSave technology on servers based on 2nd generation Intel Xeon Scalable processors with the built-in AI acceleration of Intel DL Boost.
The deep learning video provider highlighted that more than 75-80% of video streaming total cost of ownership can be attributed to distribution.
With the improved bitrate efficiency of BitSave, the company claims that its customers can significantly reduce the total cost of streaming with a bitrate saving equating to a cost-saving of $176 per hour per 5000 streams. Customers can also expect to have the same higher-quality content delivered at the same total cost of ownership.
