-
-
Comparison Picture Given vs Predicted Defect
-
Picture vs Predicted Defect (1)
-
Picture vs Predicted Defect (2)
Inspiration
We took inspiration from a Kaggle featured competition. The original competition aim was to identify defects in steel and categorise them.
Steel is one of the most important materials in our world. We used it to build our houses, to manufacture cars and many other consumer goods. If we can create better steel, we can improve our world!
The better we detect steel defects, the better we can deal with them. For companies who use steel extensively in the manufacturing process, being able to detect defects will ensure they can consistently produce high-quality products.
What it does
Our neural network aims to identify defects in steel
How we built it
Transfer learning with EfficientNet
What we learned
Image segmentation, transfer learning, collaboration, Convolutional Neural Networks
What's next for Steel Defect Detection
Categorization of different forms of defects in steel
Built With
- efficientnet
- googlecolab
- keras
- tensorflow
Log in or sign up for Devpost to join the conversation.