As AI has become pervasive, many applications in consumer to industrial grade products demand these techniques within a larger design context. The fast track deployment, improvement, and exchange of neural networks, often in the image processing space, are an important provision by architectures and so is the ease of use. AMD Vitis AI is there to help with accelerating AI inference on AMD adaptive SoCs. It provides operates optimized IP (Intellectual Property), supporting tools, libraries, models, and reference designs and allows a close integration of AI inference with the rest of your designs
To give an introductory insight into the toolchain and features, we will touch on:
Machine Learning and Deep Learning with neural networks
Versal adaptive SoC architecture and AI Engines
Introducing the NPU (DPU) into a Versal platform design
Vitis AI: quantization and compilation
VART deployment of model inference
