As the quantity of content available online is exploding, metadata is now the key to offering better user experience and securing better exposure of your news and sport content. Properly indexed, footage can be found, navigated through and reused more easily. It is still the same content, but it is far more valuable.
The importance of metadata is even more significant now that video consumption is more and more interactive and takes place simultaneously on TV, various social networks, online services and mobile applications.
Yet, creating proper metadata is not that easy. It requires a lot of manpower, is not a glamorous job and as a result, it’s really expensive.
That’s where AI comes into play. The advent of Deep Neural Networks and Convolutional Neural Networks brings ideal frameworks for video indexation. Learning how to identify people, objects and actions within video footage is no longer a research subject – it’s something the industry knows how to deliver.
As we observe it with our customers, those who invest in proper video indexing – either manual or automated – see more user engagement and better synergies between various platforms.
This leads us to foresee one thing: the pioneers who start mastering Artificial Intelligence today will be the leading media companies of tomorrow.
The good news is that access to technologies based on deep learning is no longer exclusively reserved to large corporations. Face and object recognition are available for your company and you can either train your network by yourself or ask a specialised technology partner to do it according to your requirements.
Moreover, the common belief that implementing AI-based technologies such as face recognition in a workflow requires using the cloud is not true at all – to many customers’ surprise, we can install a good system directly on their infrastructure. In fact, we can even train systems to solve very specific problems like, for example, recognising racing helmets instead of the drivers’ faces, which are hardly visible in the streaming. It is custom recognition, not pre-trained sets, that makes AI a revolutionary technology.
This doesn’t mean that cloud technology will be entirely meaningless to the process. Training an AI-based model will require huge amounts of computing power and cloud IaaS offers are very handy in this phase. However, in most cases, you won’t need to do that yourself.
So, once you’ve realised that Artificial Intelligence is already here and that it is more accessible than you might have expected, it’s time for you and your company to get to know it better and to start using it as soon as possible.
IBC Content Everywhere will be a perfect occasion for you to meet companies with the right knowledge and hands-on experience in building efficient AI-based systems for face and object recognition. With such partners, you’ll be able to progressively adopt AI and turn it into a strategic asset of your company.
Frank Coppola is Chief Executive at Hexaglobe
Hexaglobe will be exhibiting at IBC2018 on Stand 14.A16