Defect Detection Neural Network
Skills: Python (Tensorflow/Keras)
Description: Taking inspiration from published work, I implemented and tuned a convolutional neural network capable of detecting minute defects within material textures. I used Tensorflow as the backend foundation for the neural network and code in Python using the high level Keras library. During the training of the network, I utilized virtual machines through Microsoft Azure, allowing for quick and convenient model training using state-of-the-art GPUs. At the conclusion of this project, I ultimately design a novel neural network, which I title the directed segmentation-detection network, capable of identifing subtle defects with high accuracy.
Source: The source for this project is not available as this was a private work project completed during an internship. Sorry!