Abstract: 

Deep learning is revolutionizing industrial inspection with potential to achieve or exceed human level performance in certain inspection tasks and more consistent outputs when image quality varies. Generally, the computer vision models used in industrial inspection applications utilize convolutional neural networks based on state-of-the-art model architectures trained with GPU-accelerated supervised learning approaches. However, within this paradigm there is significant variety in model development and deployment, due to application constraints such as need for real-time remote edge processing or challenges in acquiring or annotating data. This presentation will share case studies highlighting technical challenges and key learnings from multiple implementations of computer vision to solve industrial inspection problems in areas such as additive manufacturing, automotive castings manufacturing, and jet engine maintenance.

Bio: 

Jeff Potts is the Advanced Analytics Leader for Baker Hughes. Jeff leads an interdisciplinary team of senior data scientists and technologists to deliver new analytics solutions leveraging AI and ML across Baker Hughes product companies, with focus on applications for industrial inspection, additive manufacturing, and energy transition.

Jeff holds a PhD in Materials Engineering from the University of Texas at Austin and a bachelor’s degree in Mechanical Engineering from Oklahoma State University. He has published over twenty technical publications and filed over ten patent applications, with technical and leadership experience in a variety of areas including artificial intelligence, energy storage, augmented/mixed/virtual reality technologies and materials development.