Edge computing has gained increasing importance in the development of Artificial Intelligence (AI) applications, enabling the deployment of intelligent applications closer to the data source and improving efficiency while reducing latency. As a result, several machine learning edge computing development kits have emerged, each with its own unique toolflow for performance evaluation. These toolchains allow developers to assess and optimize their AI applications for maximum performance in resource-constrained edge environments. In this report, the advancements in AI edge computing toolchains are explored, with insights provided on the top players in the industry. The focus is on prime examples such as Xilinx Vitis AI, Intel OpenVINO, and Nvidia Jetson-Deepstream SDK. A toolflow is proposed for the NeuroSoC project, which adheres to the fundamental principles introduced by these tools while considering the constraints of the NeuroSoC chip.

This document contains information, which is proprietary to the NeuroSoC Consortium. Neither this document nor the information contained herein shall be used, duplicated or communicated by any means to any third party, in whole or in parts, except with prior written consent of the NeuroSoC consortium. It does not necessarily reflect the opinion of the European Union.