A French startup has developed its own algorithm to visually identify objects. Rather than choosing the usual categories such as faces or animals, Prevision.io has decided to train its artificial intelligence on a very French specialty: cheeses.
You will also be interested
[EN VIDÉO] Vertical farms optimized thanks to artificial intelligence Plenty, a start-up based in San Francisco (United States), is developing a concept of a vertical farm whose production is optimized by artificial intelligence. This adjusts temperature, water supply and lighting for maximum efficiency. Because with the increase of the world population, the production of fruits and vegetables will also have to increase, if we want to feed the whole planet healthily. © Plenty
What could be more French than a debate around cheese? It is therefore no great surprise that we discover that one of the projects of the French startup Prevision.io, specialist in machine learning, is an artificial intelligence capable of recognizing cheese just from a photo. The developers called their AI Cheezam, with reference to Shazam, l’application which identifies the songs.
The visual recognition is a very complex task. While an app that recognizes cheese may seem trivial, it is a real challenge for artificial intelligence which required training on a large database containing photos of cheeses with their identification. It is not moreover the work of Prevision.io only, since Cheezam was developed in partnership with a second French startup called Isahit.
A simple photo is enough to identify a cheese
Cheezam is available as a web application, in alpha version. Just go to the site Cheezam.fr from his mobile and click on “Take a Photo”. Take a new photo of a cheese or select one already saved on the device. The AI automatically tries to identify it and indicates its confidence index. In case of error, it is possible to send the image indicating the name of the cheese in order to improve the accuracy of the AI.
For those who would like to integrate Cheezam into their own application, the site offers the code to call their API in JavaScript, Node (Axios), Python, Curl et Swift.
Interested in what you just read?
.
