PARIS—With the speed of manufacturing being what it is in 2018, the level of attention humans can maintain for hours at a time is a problem—for which there are solutions. IBM’s cognitive assistant is one. Scortex’s two-year-old Quality Intelligence Solution is another.
Scortex CEO Aymeric de Pontbriand commonly relays this tale from one of his first customer’s experience with Quality Intelligence Solution: During the final test of Scortex’s software at that French automotive supplier—when the customer ran the software in parallel with its human-inspection system—the quality expert “came with a batch of ‘good’ parts and a batch of ‘defective’ parts that her inspector provided to her,” he said. “During the test, our machine detected a defect in one of the ‘good’ parts, and the expert was like, ‘Oh, your machine was wrong.’ But when she inspected the part, she found a defect.”
Make no mistake: The automotive supplier trains inspectors for two weeks before they are allowed to perform quality control work on its production line. “There is expertise in being able to inspect,” de Pontbriand said. “But, of course, over time and over the course of the day, the level of attention that you can provide will decrease.”
Paris-based Scortex works to automate complex visual inspection with its in-house developed hardware—which involves a processing chip connected to cameras—and data-processing software.
“The hardware part is more than a computer system,” he said, noting that “the box” allows for input and output with cameras and other sensors and communication with the production line. The box also enables “real time processing so that the processing time is never a bottleneck for the production cycle.”
The deep learning capabilities of the computer vision pipeline Scortex offers let its customers “tackle visual inspection that no other standard solution can tackle,” such as glossy plastics components, said de Pontbriand, 28.
He founded Scortex in 2016 along with CTO is Serge Zloto, COO Christophe Raix and hardware lead Nathanael Hania.
Today, it employs about 25 people, and de Pontbriand plans to double that number 18 months from now.
The “s” in Scortex comes from the Greek letter sigma, he said. “It’s the primary element of artificial neural units, and we use mainly artificial neural networks.” The rest of the name comes from the word cortex, the part of the brain that does pattern recognition.
Artificial intelligence (AI) is involved in Scortex’s offering on two levels.
“First, we developed a specific pipeline using deep learning that allows us to extract information about the quality from images,” said de Pontbriand. “You can find software that allows for basic visual inspection, but when it comes to very complex visual inspection tasks, usually it still requires humans. So we developed an algorithm that lets us recognize small defects and unusual patterns.”
Second, Scortex collects a lot of information about quality trends. “This enables the time series analysis we do to detect anomalies and help find trends in terms of quality,” he added.
Scortex’s offering has attracted €1.8 million ($2 million), from British and French investors.
The company is targeting the automotive space, “where quality is really critical, especially for supplies”—and where a lot of visual inspection is still done manually, de Pontbriand said.
Automakers appreciate the machine learning aspect of Scortex’s product because “it keeps learning over time”—from continuously collected, real-time production data.
“That means you can understand what your defect rate was for the day,” he said. And manufacturers can easily see what type of defects they are experiencing, as well as where the defects are occurring on their products. On top of that, with a connected solution, “you can compare different factories and understand the trends and make sure that the best practice in terms of quality is shared across the organization,” he added.
De Pontbriand last week met with Business France here to prepare for visits to Detroit and Toronto.
“We are targeting customers that have operations all over the world. And, of course, North America is a pretty important region when it comes to automotive, especially the Detroit region.”