Presented By O'Reilly and Cloudera
Make Data Work
March 13–14, 2017: Training
March 14–16, 2017: Tutorials & Conference
San Jose, CA

Deep learning in the automotive industry: Applications and tools

Andre Luckow (BMW Group)
1:30pm2:00pm Tuesday, March 14, 2017
Data science & advanced analytics, DTL, Strata Business Summit
Location: LL20 B Level: Intermediate
Secondary topics:  Deep learning
Average rating: *****
(5.00, 1 rating)

Deep learning has been proven to be very effective in domains such as computer vision and natural language understanding, and an increasing number of automotive applications can benefit from deep learning. Andre Luckow offers an overview of use cases for deep learning in the automotive industry, surveying the current state of the art in libraries, tools, and infrastructures (e.g., GPUs and clouds) for implementing, training, and deploying deep neural networks.

Andre focuses on convolutional neural networks for computer vision use cases, such as the visual inspection process in manufacturing plants, discussing his experience developing an end-to-end deep learning application utilizing a mobile app for data collection and process support and an Amazon-based cloud backend for storage and training. A particular challenge is the availability of training data in particular for enterprise applications and the associated cold-start problem.

Andre describes the creation of an automotive dataset that allows us to learn and automatically recognize different vehicle properties. Cloud and on-premises infrastructures (including multiple GPUs) in conjunction with different neural network architectures (AlexNet, GoogleNet) and frameworks (TensorFlow, Caffe, and Torch) are used to evaluate the dataset. Training and inference performance as well as the accuracy of the classifier are then assessed before the classifier is evaluated during real-world use within the manufacturing process.

Photo of Andre Luckow

Andre Luckow

BMW Group

Andre Luckow is a project manager and researcher at the BMW IT Research Center in Greenville, South Carolina, where his work focuses on interdisciplinary research and applications at the intersection of data infrastructure, data science, and machine learning in the automotive domain. His specialty is the application of computing technologies to problems in business and science bridging cross-functional gaps to create value via process improvements or the enablement of new types of products. He is particularly interested in deep learning applications and system-level challenges related to deep learning, streaming, and edge computing. Previously, Andre served in a number of positions at BMW Group IT in Munich, Germany. He holds a PhD in the field of distributed computing from the University of Potsdam, Germany.