With the application of Machine Learning becoming a common task in consumer appliances as well as in common equipment in the factory field, easily adding such features into advanced products becomes a key element of modern system design flows.
With this course, we utilize Vitis™ AI as a concept to apply own neural network solutions within the FPGAs and embedded architectures of AMD adaptive SoCs. A hands-on review of the basics of neural networks in image processing determines the benefits of inference acceleration. Choosing an application focused approach to derive the requirements of complex neural networks, the acceleration of such massive computation is achieved by means of the Deep Learning Processing Unit (DPU) as a common IP target architecture for any AMD Adaptive SoC family. To map out trained CNNs on platforms with DPUs the Vitis™ AI toolkit is introduced as a powerful tool to translate from various ML toolkits such as Py Torch and TensorFlow. This mapping process quantizes the original float data into efficient fix point representations as used in the DPUs. To run the model operators on the target, the AI compiler creates an executable that can becalled from any host application, which may be programmed in Python or C/C++.
To give an overview of the wide range of readily available deployment variants, we will work with the Vitis™ AI library and discuss the supporting pre- and post-processing approaches. Such computational blocks can also be mapped in hardware or computed within the embedded application itself. A final look into the generation of custom platforms using the DPU provides further insight to give a starting point for the own designs.
10/14/2025 - 10/15/2025 Time Zone : (GMT+01:00) Amsterdam, Berlin, Bern, Rome, Stockholm, Vienna Seats Remaining : 12 Venue : DEU, Stuttgart - TBD PLC2 Venue Address : TBD,Stuttgart,GERMANY
12/1/2025 - 12/2/2025 Time Zone : (GMT+01:00) Amsterdam, Berlin, Bern, Rome, Stockholm, Vienna Seats Remaining : 12 Venue : DEU, Munchen - TBD PLC2 Venue Address : TBD,Munchen,GERMANY
