“We are unveiling the first MCU AI developer cloud, which works hand-in-glove with our STM32Cube.AI ecosystem,” said ST MCU v-p Ricardo De Sa Earp. “This tool brings the possibility to remotely benchmark models on STM32 hardware through the cloud.”
STM32Cube.AI is the existing desktop version, which can generate STM32 AI libraries from trained neural networks, and can now act as a front end to the new tool: STM32Cube.AI Developer Cloud, which has:
- An on-line interface to generate C-code for STM32s without prior software installation.
- Access to a GitHub-based repository (‘model zoo’) of trainable deep-learning models and demos to speed application development – Initially: human motion sensing, image classification, vision-based object detection and audio event detection.
- On-line benchmarking service for edge-AI neural networks on STM32 boards.
STM32Cube.AI Developer Cloud access here (MyST registration requited)
The tool has been undergoing testing and evaluation by several embedded development customers.
