How did the creation and development of Scan Ai come about?
Emmanuel Jalenques: In 2019, Interscience R&D engineers began work on the use of CNNs applied to colony counting. Our production since 2006 of automatic colony counters with conventional algorithms provided a good knowledge of customers’ use cases.
The subsequent development of the ScanStation, a real-time incubation and colony counting station, in 2017 made it possible to automatically generate approximately 1 million correctly annotated Petri dish images. We therefore had a solid and very varied database of analyses and plates.
Did you collaborate with other research units for this development with AI?
EJ: In 2021, Interscience has formed several partnerships with university image processing laboratories, as well as with engineering schools, to accelerate the theoretical development of AI applied to microbiology.
At the same time, a user test campaign was carried out at several Interscience customers, including Merieux Nutriscience in Merville, one of the world’s largest service laboratories, to test and validate performance in real-life use cases.