Utilization of Intel's OpenVINO Computer Vision Toolkit

Research Question – How can Deep Learning processes be used to analyze and detect defects in manufacturing equipment in industrial factories around the globe?  Methods – Using TensorFlow graphs, Machine Learning Training on Google Cloud, and Intel’s OpenVINO Model Optimizer, deploy a trained neural network model to detect defects on a live video stream.  Results – Using about 800 pictures, a trained model can detect defects in an item with up to 99% accuracy. Certain camera angles and lighting may sometimes affect these results.  Conclusions – Deep Learning can create very accurate results with only a few hundred pictures used for training. With an abundance of data/images, models will be able to detect factory defects at any viewpoint and in any environment, allowing full automation for quality assurance processes.