Deep learning is a set of automatic learning methods that abstractly models data through articulated architectures of different non-linear transformations. These techniques have enabled significant and rapid advances in the fields of sound and visual signal analysis, in particul ar:
• Facial recognition,
• Voice recognition,
• Of computer vision &
• Of automated language processing.
With the trivialization of cameras, the success of cam action, drones and hosting platforms, the number of video data has exploded over the last ten years. In addition, the company has a myriad of devices that continuously record our daily lifestyle habits and behaviors: C CTV installations, smart cameras, …
These masses of data remain difficult to exploit since their analysis is time-consuming making it impossible by traditional means. The solution to this problem is automation, a method which has made rapid progress with the rise of deep learning.
In partnership with local universities and specialized research departments, the ESI R & D team has decided to integrate deep learning functions into its products.
The first results can be found in the AV1 module for automated video
• The face detection in the image,
• The classification of objects detected in the image
• The search for events from very long accelerated films
(Impossible to efficiently do by a human without assistance of an artificial intelligence)
Numerous applications allowing Cyberguardian support will be launched in the coming months on an experimental as well as operational basis.