Descripción del proyecto
Reducir el tráfico de datos de vídeo para mejorar la calidad
Desde los días de gloria de los videoclubs y las cintas VHS, la forma de disfrutar del cine en casa ha cambiado mucho. Hoy en día, los vídeos son de transmisión libe («over-the-top», OTT), por lo que las emisiones en directo son una de las principales fuentes de tráfico en internet. El mercado OTT crece con rapidez, impulsado principalmente por Netflix y YouTube. Pero el auge de las emisiones en directo se da sobre todo en los países desarrollados. En el resto del mundo, las situación es diferente. Por ejemplo, la mitad de la población mundial aún no tiene acceso a internet. En este contexto, el proyecto ENHANCEplayer, financiado con fondos europeos, pretende reducir la brecha digital disminuyendo el tráfico de emisiones en directo en un 80 %, pero manteniendo la resolución del vídeo y aumentando la calidad. En concreto, el equipo aprovechará el aprendizaje profundo para aumentar automáticamente la resolución y la calidad de los vídeos en los dispositivos de consumo.
Objetivo
Video streaming consists in the encoding, transcoding, delivering, decoding & displaying of videos on smartphones, tablets, PCs, smart TVs & consoles. Videos distributed “over-the-top” (OTT) are streamed rather than distributed through traditional networks (broadcast, pay TV) and distributors (cable, satellite).
OTT is a vibrant $25bn market (2016) poised for spectacular growth: by 2025, it will reach $129bn, a 5.2x increase driven by powerhouses like Netflix and YouTube. Habits are changing fast: by 2020, 32% of videos will be consumed OTT (16h/week), vs. 5% in 2000 (2h/week).
Video streaming represents 63% of Internet traffic (2018) and is on the rise, going from 76k petabytes per month in 2018 to 160k in 2021 (x2.1) and driven by a push to 4K that quadruples traffic. While developed countries are challenged to address this surge, half of the World’s population still doesn’t have the Internet and is unlikely to be able to stream content when they get it, reinforcing the “Digital Divide”.
Reducing this burden will lead to a faster, cheaper and more stable Internet. It will slash content delivery costs, which reach millions of dollars a month for media companies. It will result in major savings for Internet providers that invest ~$300bn/year in infrastructure, or give them the opportunity to achieve much more for a similar investment.
Our goal is to decrease video streaming traffic by 80% while retaining video resolution and increasing quality. To do so, Artomatix and THEO will develop ENHANCEplayer, a solution that leverages Deep Learning to automatically increase the resolution and quality of videos on consumer devices. Built on top of THEO’s successful THEOplayer, it will consume the data of, say, a 360p video but display a High Definition one (1080p), effectively saving 84% in traffic. Public broadcasters NPO, RTP & VRT will help Artomatix & THEO define the product specifications, provide them with test videos and test the product once it is developed.
Ámbito científico
- natural sciencescomputer and information sciencesinternet
- engineering and technologymechanical engineeringvehicle engineeringaerospace engineeringsatellite technology
- engineering and technologyelectrical engineering, electronic engineering, information engineeringinformation engineeringtelecommunicationsmobile phones
- natural sciencescomputer and information sciencesartificial intelligencemachine learningdeep learning
Programa(s)
Régimen de financiación
IA - Innovation actionCoordinador
DUBLIN 4 Dublin
Irlanda
Organización definida por ella misma como pequeña y mediana empresa (pyme) en el momento de la firma del acuerdo de subvención.