Cerrion launches video-based process monitoring solution

While many production lines are equipped with sensors to ensure smooth operation, many other manufacturing processes still lack real-time visibility into daily deviations. Cerrion closes this gap with its newly launched computer vision and AI-based solution, and first paying customers in four different countries are already testing the tool.

Deviations in a manufacturing process can lead to inefficiencies and unplanned losses. Container glass manufacturing delivers an ideal example of a complex process that often deviates from its stable operating point. However, manufacturing companies in this sector still lack real-time insights into the timing and magnitude of these deviations and automatic mitigation mechanisms—failure to address process deviations promptly could result in extended downtime, machine damages and high-unplanned costs.

Responding to this challenge, Zurich-based deep tech startup Cerrion is launching a new solution based on computer vision and artificial intelligence research. Their product is a software/hardware solution. It comprises lightweight camera sensors that produce anonymized video streams and an accompanying AI-based software component, which analyses the video observations of an operation to detect abnormal events. In case of a deviation or anomaly, relevant stakeholders will receive an alert in real-time. The software is licensed under a SaaS model, while the hardware is a one-off investment.

“After comprehensive market research, we concluded that the problem we are tackling is a big challenge. Our goal is to monitor every manufacturing process where deviations are visually distinguishable. Our product relies on self-supervision to train state-of-the-art deep neural networks and novelty detection to detect deviations automatically. This allows deploying our product to new processes without annotating hours of data as in conventional approaches”, says Karim Saleh, Co-founder of Cerrion.

Cerrion is currently working with leading and renowned multi-million and billion-dollar revenue manufacturing companies in the DACH region and beyond. Based on the initial research and Proof of Concepts (PoCs) conducted with their industry partners, the team has discovered multiple use cases – these are currently being validated in further pilot projects with enterprise and medium manufacturing companies in different sectors. With the go-live, the startup is now tapping into a $12 trillion industry to serve existing and new customers across Europe.

Entrepreneurial and deeptech knowhow
The three founders have entrepreneurial experience and bring a wealth of expertise in computer vision, engineering, and artificial intelligence. Karim Saleh has a degree in engineering from ETH Zurich and has previously led Unicorn Labs to help startups reach a minimal viable product faster. In 2020, he was recognized by Forbes in their annual 30 under 30. With a PhD in computer vision from ETH Zurich, Michael Gygli is an AI expert. He has previously worked at Google as an AI researcher and has published more than 20 research papers in Computer Vision and AI. Nikolay Kobyshev is a serial entrepreneur who has previously co-founded Europe’s largest hackathon, Hack Zurich, the AI startups Spectando and Assaia. He also holds a PhD in computer vision from ETH Zurich. With a growing demand for their solution, the startup is currently growing its team and looking for additional talent to join them.