Traditional inspection protocols already deliver outstanding results but at the expense of manual support. B.R.AI.N automates these complex processes and improves the inspection performance at the same time. B.R.AI.N works on the basis of Deep Learning models. These models are trained to make best possible and reliable inspection decisions.

With B.R.AI.N, you benefit from higher detection rates and reduced false rejection rates (FRR) of up to 99%. Avoid costly retesting (e.g., with additional semi-automatic testing machines and specialized visual inspector) and maximize efficiency and output quality.

B.R.AI.N is a proven and fully compliant supervised deep learning application. It is trained based on your inspection images and is, therefore, completely transparent and always under your control. B.R.AI.N extends your existing recipes with AI capabilities without having to develop new recipes. Future adaptations and updates are also easily possible without the need for re-certification.

To build your customized B.R.AI.N extension, we follow a proven and standardized 6-step process that guarantees the best results even under high-performance production conditions.

• Initialization
  Defining the scope based on the ejection analysis and extracting the test data
• Model development
  Pixel-perfect categorization and assignment of criteria to identify the parts to be inspected
• Model integration
  Applying data science to train, test, and validate the classification model
• Implementation
  Commissioning and fine tuning of the trained AI and installation of edge devices, if required.
• Model monitoring
  Fluctuations in production are not preventable. We give you the right tools to analyze current model behavior
• Model finetuning
  B.R.AI.N. enables existing models to be adapted to new requirements. All supported by a fully compliant system.

No two pharmaceutical production lines are the same. Our experts examine your inspection situation to identify opportunities to optimize inspection performance, down to a single camera. Based on the results, we develop a customized solution that perfectly combines traditional and B.R.AI.N capabilities.

While the overall scope of the project depends on several variables, such as machine AI readiness, the number of recipes to be optimized, and the number of camera stations, the implementation itself... is fast. Within a few days and with minimal downtime, B.R.AI.N is implemented and running in full production mode.

• New inspection machines AI ready
  Extent existing recipes with B.R.AI.N right from the start and without any hardware installation.
• Retrofit existing older machines
  If the existing machine has no computing capability, you can benefit from external B.R.AI.N. edge devices. The machine becomes AI-capable without significant production interruptions or qualification effort for existing recipes. Machines with SVIM without sufficient computing power are upgraded with a B.R.AI.N GPU chip.