Fake Realities, Digital Owls and Maximized DL

Published: 07/18/2019  

Last Updated: 07/18/2019

AI Innovation Highlights from the AIDC Summit in Munich on July 9th

Welcome talk at AIDC Munich

Fooling the eye with Deepfakes, preserving endangered species with AI-powered drones, and getting the most from deep learning on Intel® architecture – these innovations and more were highlighted during the Intel® AIDC Summit in Munich, Germany, on July 9.

Offered by invitation only to small groups of enterprise developers, the AIDC Summit series spotlights the achievements of leading and upcoming AI innovators and Intel partners. This all-day conference provided hands-on training in how to apply the Intel® AI portfolio to create the ideal hardware/software system for deep learning applications.

The Summit in Munich featured these noted AI developers and their demo presentations:

Deepfakes 2.0 by TNG Technology Consulting addressed how neural networks are changing the world, as illustrated by the emergence of deepfake technology. Deepfake (a combination of “deep learning” and “fake”) applies the use of machine learning for human image synthesis to superimpose one person’s features and expressions onto source images of a different individual – and to do so convincingly. As with other audio-video technology tools, the software and compute resources necessary for deepfakes are increasingly available to the individual enthusiast.

Deepfakes source video vs destination video sample

The Digital Owl Project from Fujitsu deploys AI-powered drones to help preserve endangered species in Australia. Using AI video analytics, Intel and Fujitsu replaced manned helicopter expeditions with remote operated drones and used video analytics to identify endangered species and pest species. This solution lowered project costs and reduced helicopter fuel emissions, allowing for more frequent surveys.

Computational Flow in Digital Owl

Maximizing Deep Learning Training on Intel® Architecture featured a two-hour hands-on session showing techniques for deriving peak performance when training deep learning models on Intel® architecture. Participants examined the benefits of using the Intel AI portfolio of processors and optimized software for accelerating deep learning model training on the latest generation of Intel® Xeon® Scalable processors. The result was a trained model using Intel® Optimization for TensorFlow*, which was then used in a second hands-on session. The workshop material is available as an online course: AI from the Data Center to the Edge – an Optimized Path Using Intel® Architecture.

AI on Intel

Efficient Model Deployment using Intel® Distribution of OpenVINO™ Toolkit demonstrated its ease of use to optimize and deploy a trained model on a range of hardware. The course explored sample AI solutions that come packaged with Intel Distribution of OpenVINO toolkit to see how they can be leveraged for quick prototype building.

Join Us for a Day of Hands-on AI Learning

The invitation list for every Summit consists of data scientists, researchers, developers and management-level IT staff. Attendees should have an understanding of AI principles, machine learning and deep learning, Python* coding experience, and familiarity with TensorFlow*, Caffe* and other frameworks. For further information, or a refresher, Intel recommends introductory courses covering Machine LearningDeep Learning and Applied Deep Learning with TensorFlow.

AIDC Intel Summit Series 2019 Logo

Find an Event Near You and Register

The AIDC Summit Series is being hosted in cities worldwide. These complimentary events have limited seating – make sure to check out the Intel AIDC Summit Series website to find a location near you.  

Want to learn more?

For details on the Intel AIDC Summit Series content, training materials and other Intel AI resources, check out the  Intel® AI Academy.


Product and Performance Information


Performance varies by use, configuration and other factors. Learn more at www.Intel.com/PerformanceIndex.