Case Study

Check the quality of frozen food 15 times faster with AI

A leading food manufacturer relies on Computer Vision to inspect vegetables for visual defects in real time. With a accuracy of 95 percent quality control is now not only significantly more reliable, but now takes only 1 minute per ton of vegetables instead of 15 minutes.

Contact us

PRIVACY

Je déclare avoir lu et pleinement compris la note d'information sur la protection des données personnelles Privacy Notice et j'exprime par la présente mon consentement au traitement de mes données personnelles par Reply SpA à des fins de marketing, en particulier pour recevoir des communications promotionnelles et commerciales ou des informations concernant des événements ou des webinaires de l'entreprise, en utilisant des moyens de contact automatisés (par exemple, SMS, MMS, fax, e-mail et applications web) ou des méthodes traditionnelles (par exemple, appels téléphoniques et courrier papier).

Something went wrong with the form submit. Try again.

#computervision
#artificialintelligence
#quality
#food

The project at a glance

Freshly harvested and checked by AI: how today's food industry ensures the perfect taste experience

The scenario

Excellent color, shape, and quality

Whether it's green peas, strawberries, or broccoli, consumers expect frozen foods to be in perfect condition: vibrant in color, consistent in shape, and free from any foreign objects. While this may seem like a given, ensuring such quality is a complex logistical challenge. In the EU, approximately 9 billion strawberries and over 7 trillion green peas are harvested - and inspected - each year.

For one of Europe’s leading frozen food manufacturers, maintaining this level of quality has become increasingly difficult. Until now, quality control was performed manually on a sampling basis - a time-consuming, error-prone process reliant on skilled personnel. The growing shortage of qualified workers only added to the strain.

The solution

Fully automated visual inspection using AI and Computer Vision

In collaboration with Machine Learning Reply, the food producer implemented an AI-powered system for fully automated visual inspection directly on the production line. High-resolution cameras continuously monitor the passing produce - every single pea, every single berry. A custom-trained Computer Vision solution accurately detects visual defects and foreign objects, transmitting the results in real time to a centralized dashboard.

This advanced solution not only replaces manual random sampling, but also enables 100% real-time inspection, ensuring consistently high product quality and full transparency throughout the production chain.

How we did it

Real-time processing with adaptive image optimization

Consistent image quality is critical for accurate and reliable inspection results. However, varying lighting conditions, production speeds, and product positions can compromise image clarity. To address this, the experts developed a custom algorithm that dynamically adjusts exposure, lighting, and capture timing, systematically eliminating common issues such as motion blur.

Powerful NVIDIA edge devices, combined with DeepStream technology, enable real-time processing and immediate application of insights. Close collaboration with the food manufacturer led to a practical, user-friendly system that requires no specialized AI expertise, ensuring seamless integration into existing operations.

The results

Greater precision, less off-sort

The introduction of the AI-supported quality assurance system brings measurable benefits.

15 times faster

inspection thanks to real-time processing

95 % detection accuracy

for product defects and foreign objects

Reduced scrap

thanks to precise error detection before further processing

Full transparency

regarding inventory and product quality

Future-proof

thanks to automated retraining platform for new product variants

Machine Learning Reply

Machine Learning Reply offers customized end-to-end Data Science solutions, covering the entire project life cycle – from initial strategy consulting, data architecture and infrastructure topics, to processing data with quality assurance using Machine Learning algorithms. Machine Learning Reply has extensive expertise in the field of Data Science in all key industries of German HDAX companies. Machine Learning Reply empowers clients to successfully introduce new data-driven business models and to optimize existing processes and products – with a focus on distributed open source and cloud technologies. With the Machine Learning Incubator, the company offers a program to train the next generation of decision-makers, data scientists and engineers.