The Chainless - DeepVA
Artificial intelligence AI - Machine learning ML
DeepVA is an intuitive data mining platform that automatically receives training data, customizes AI and then uses it to extract and analyse any type of information from images, videos and livestreams. DeepVA is a platform for the media industry and for capturing and managing all large archives. It has an intuitive interface and is easy to integrate with existing systems. DeepVA can be used on premise, and also has an API that works across all technologies.
Developed on the basis of current research, DeepVA is performant, scalable and highly accurate. Its pre-trained AI models have been developed in close collaboration with the media industry and strive for accuracy without compromising on speed. We keep an eye on the changing environment and continuously design flexible AI solutions for the upcoming challenges in the industry. In addition to pre-trained models, DeepVA allows you to train your own AI models by classifying new data with supervised information when you have only a few training samples.
When pre-trained models are unable to identify unknown individuals, such as new celebrities or local athletes, DeepVA provides the ability to integrate them into a custom AI model. These custom models can be used to analyze nearly any video footage. For providers of large image and video archives, this results in very large savings and optimization potentials in daily work processes. DeepVA makes it easier for companies to take the first step towards AI, which will be indispensable in the near future.
Today, visual data, such as images, videos and live streams, make up the majority of the media used, accounting for 80% of online traffic. The production and consumption of this data is rapidly increasing from year to year. This explosion of information makes it almost impossible for a company to structure, analyse or retrieve this data. Our AI platform takes care of these tasks and allows you to extract information from visual data in a completely automated, cost-effective and time-efficient way to gain insighs and make important decisions.
Moreover, AI is ubiquitous and will become very important in the future. Many companies are unfamiliar in this field, but must follow suit in order to remain competitive. The introduction of a proprietary AI model is therefore difficult and costly. Finding suitable experts costs time and money and is risky because it is unclear whether the AI will deliver the desired effect. This is why many companies are hesitant to implement AI. Our product therefore offers the possibility to create AI models in the field of computer vision in a user-friendly, uncomplicated way.
DeepVA allows companies to easily integrate our AI technology into their workflows and enables them to develop their own AI models to find very customized solutions. In addition to integrations at large broadcasters and city archives, DeepVA can be seamlessly integrated into the infrastructure of larger data centres via an easy-to-configure interface (API) and extends existing software, such as media asset management or digital asset management systems, with new AI functionality. Example of an integration is with Arvato System's implementation of VidiNet, a cloud-based media service platform of the Arvato Systems brand Vidispine.
The Vidispine team wanted to incorporate AI tools into the users' environment, allowing users to independently monitor and control the recognition of entities, create their own AI models and QC their training data. The integration of DeepVA in VidiNet enables the creation of individual AI models in the company's own MAM system without any prior technical knowledge and just one image to train a new face model. DeepVA covers a large range of the total AI application areas such as Visual Concept Recognition, Face Recognition, Landmark Recognition, Brand/Logo Recognition, and Text Recognition. The software runs on the cloud at Vidispine and is the ideal complement for VidiNet users, as it facilitates the move towards AI, which will be essential in the near future.
DeepVA is a video analysis system consisting of software and hardware. The software is based on Artificial Neural Networks and provides users with an extensive, intelligent database. DeepVA regularly expands this database with new object classes, enabling it to continuously increase its analysis capabilities. The video analysis is composed of different modules, which together are able to check a digital video for many different features including object recognition, face recognition, text recognition, and store the resulting metadata in the database.
The hardware is an optimized, scalable, redundant server cluster that enables real-time, faster analysis of the video data. DeepVA is the deepest video analysis for TV broadcasters, agencies, film production companies, streaming services, publishers and archives. With an innovative and unique program core (CoreAI), we enable our system to learn independently and improve continuously. New modules extend DeepVA and permanently increase the depth of analysis. DeepVA aims to revolutionize the way we deal with media and make information accessible through intelligent software.