Cyberus

Inspiration

Profit time and cost in the chip production line by using various machine learning methods.

What it does

Project takes the images of the chips and detect if it has a defect or not with a high precision.

How we built it

We have created an Azure database for PostgreSQL and uploaded both raw and preprocessed images to use in the model training. We have applied 5 different preprocessing methods to the raw images in order to increase the accuracy. We have created a Custom Vision project on Azure to train our model.

Challenges we ran into

It was hard to create a fully connected cloud-based project.

Accomplishments that we're proud of

We have met with all of the requirements on time. Model works with a high precision. Somehow, all of us have implemented different parts using different technologies that we have never used before.

What we learned

We have learned how to create a fully working Microsoft Azure AI application.

What's next for Cyberus

We can apply different methods to increase the accuracy like data augmentation, we can train different models once we have a bigger dataset like CNN, RNN in the future.

Built With

• azure
• customvision
• postgresql
• python