AI acceleration through specialized hardware and software

This study examines the use of AI hardware and software in the context of computer vision based on research and concrete benchmarks on several AI platforms. It provides an overview of the accelerator hardware available on the market and classifies it into application scenarios ranging from IoT to data centers. In addition, detailed benchmarks focusing on performance and efficiency are carried out to compare several AI hardware-software platforms and discuss their advantages and disadvantages. These results provide important insights and criteria for the evaluation of AI systems. Based on this, more informed decisions can be made about the choice of AI hardware and software optimizations for accelerating AI systems in order to develop more efficient and cost-effective systems.