Technological projects portfolio


Artificial intelligence for real-time object recognition in videos

Investment: €406k

Scope: Digital and Electronics

Scientific field(s): Physics – Chemistry and Optics / Mathematics, Information and communications science and tech. (STIC) and nanotechnology

Institution(s): Télécom SudParis – CNRS

Development: Technology to be marketed

#Algorithm #Asynchronous


Image processing is now capable of recognising objects in static images.

However, no solution yet exists that can perform this recognition for video. This is due to the quantity of data involved in this problem – a video incorporates many frames, and usually many more objects than a static image.

The CBORBIG project solves the problem of real-time recognition in videos.


The algorithms developed make it possible to process huge quantities of data and to detect any object automatically.

The solution also offers the potential for producing interactive videos by inserting links into the objects identified, giving access to additional information.


This project could be targeted at several different markets:
– In the fashion industry, automatic recognition based on a photo of an item of clothing, shoes etc.
E-commerce: inserting a link to a website to buy a product identified in a video, optimising the visit/purchase conversion rate,
Advertising and digital communication: interactive videos providing information about a brand’s products through links inserted into the objects in the video. Media monitoring,
Security: video surveillance.