Heavenly insight and the need for accurately detecting faults in a manufacturing setting.

What it does

Magic, but maybe also automatically process files uploaded via Azure IoT Hub, detect and log any defects and mark them in the image with a bounding box.

How we built it

  • Azure custom vision to train and detect faults
  • Azure functions to trigger the pipeline once a new image is uploaded
  • Azure Cosmos DB and blob storage for storing files and data
  • Power BI to display relevant info to the appropriate people

Challenges we ran into

Deciding on the proper data augmentation technique and correctly hooking up the cosmos db to PowerBI

Accomplishments that we're proud of

The high accuracy of our model and great UI/UX

What we learned

The meaning of life itself and how to properly use the Azure product suite.

What's next for DeepDefectDetector

World domination. Further integration into a real-world process.

Built With

Share this project: