MSc thesis: Anomaly detection during tabletting hos Novo Nordisk

During the production of tablets several kinds of defects can happen. 

These defects can affect the visual appearance and the product qualities.

After initial results with  we suggest to investigate other algorithms or work on optimizing and further developing the current method for image recognition and classification to show how this can be used in-line in a continuous production line of tablets.

We hope to establish the optimal process and technology to support this.

Key Challenges

  • Algorithm screening
  • Algorithm optimization
  • Algorithm/Model robustness to process parameters
  • Algorithm transferability between products/sites/equipment's
  • From POC to Production
  • QA approval/Validation

Competences needed

  • Data Analytics
  • Data Engineering
  • Process SME
  • Support for prototyping/In-Line data collection and analytics
  • Automation Engineering  


Given the above unit of measurement as the key, the value is then given here. How specific this needs to be depends completely on what is possible. You will agree with the Line of Business to define both the unit of measurement, and the value of this. It can be both a number, an interval, or a short expected outcome, if it is e.g. continuation of business as the unit of measurement.

Merk: Du skal ofte ha forhåndsgodkjennelse fra universitetet ditt eller studieveileder, for å sikre deg at prosjekter eller masteroppgaver på Graduateland kan bli akseptert som en del av studiet. Kontakt det rette organ i god tid for å sikre deg at du velger det rette prosjekt.