ARGOS – recovery of stolen objects


Formed by an association of French insurance companies, ARGOS monitors valuable items placed for sale in order to spot items that may be stolen property.

The Challenge
Every year thousands of valuables are stolen – many are sold on the black market or by legitimate dealers and disappear forever. As a result large compensations are paid to insured property owners. Traditionally, rigorous investigations involving manual comparison of images were conducted to uncover stolen objects with a low success rate.

The Solution
ARGOS manages the database FIBAR (Fichier Informatique des Biens Assurés Recherchés) that contains images and data about reported stolen objects provided by the insurance company members of ARGOS. These images are compared with images from the web, dealer catalogs and images of items found in investigations by law enforcement such as the French agency OCBC (Office Central de lutte contre le trafic des Biens Culturels). To expedite these image investigations and automate part of the process, ARGOS has applied image recognition from LTU technologies. As a result, ARGOS can now, in just a few seconds, analyze hundreds of thousands of image which is particularly powerful for investigations since with an image there are no language barriers.

The Results
“In barely a few seconds, the system analyses the 16,000 images furnished by our members and compares them every day with the hundreds of thousands of images we capture on the Internet or receive in catalogues or from investigations,” explains Mr. Marsaud, Director General of ARGOS. “One of the major advantages of LTU engine is that we can consider increasing the size of the FIBAR database as well as the number of catalogues to analyze, without needing additional budget or additional human resources.”