Detecting Defects in Production & Maintenance Department in a Steel Company

• Tharun Kommaddi

• www.github.com

Designed a Data Pipeline that does Data Preprocessing, Feature Selection, and Feature Engineering.

Developed a layered Deep Learning Pipeline to automate the process of detecting and localizing defects found in Steel manufacturing by performing classification & segmentation using Convolution Neural Network, Resnet(Residual Network), Transfer Learning, Resunet.

Experimented with these Networks and got an accuracy of 87%