SI MEDIA & AI
YES! pushes with AI
AI is gaining more and more space in the world, especially in activities already covered by the media.
New technologies continue to renew themselves by bringing cutting-edge updates, and SI Media follows this flow with integrations to make YES!MAM interact both with our in-house AI and with the integration of third-party AI engines, such as whose provided by big players like AWS and Microsoft. SI Media developed its AI framework with an internal project that started in the second half of 2022 and targeted a group including TV broadcasters, media networks, production houses and media groups.
The aim was to develop a framework, seamless integrated with YES! platform to extract descriptive metadata from video and audio contents, which would then be processed and managed by YES!MAM.
A major plus of SI Media AI is that it can be deployed not only in the cloud but also on premises, allowing the Customer to keep its own contents on site. This first phase of integration and enhancement of the platform led us to be able to achieve the following results with the AI resources we have today:
Face Recognition
Digital method of personal identification already built into devices and constantly updated.
In this case, YES!AI is able to search the archive for videos in which a person appears and then see within the file in which parts he or she is present, displaying a timeline with the presence or absence of each person found, even if several people are present at the same time.
Content moderation
It allows the recognition of explicit content, nudity for example and sensitive scenes, the handling of which depends on the media and the type of images suitable to be transmitted according to the target audience.
As with facial recognition, content is identified in the archive and shown in the timeline.
Speech to text
It automatically recognises the spoken language by transcribing its content with the possibility of generating automatic translations into another language.
Not to be overlooked is the fact that it is not a real-time function but a written version of an entire audio recording.
Upcoming developments include logo recognition and object recognition: the former allows one to find logos in videos, thus useful for advertising placements (such as avoiding a competing brand’s commercial), the latter to find different categories of objects in the video, again useful for the advertising planning. To all this is added the further possibility of performing speech-to-text in real time, an update already present in SI Media roadmap.
YES!MAM is able to integrate both leading machine learning and AI services such as Amazon Web Services, Azure and Google Cloud as well as services provided by companies focused on AI such as Metaliquid and ComCast.
The results of the AI processes and the generated metadata are imported via REST API or WEB Services in YES!MAM and displayed in the UI.
Keep following and visit us at the incoming NABSHOW 2024 (boot SU1057) to find out how SI Media is focusing his efforts to enlarge AI services provided by YES! platform.
