Nuria Pazos, researcher at the University of Applied Sciences of Western Switzerland, presents a key initiative within dAIEDGE: the implementation of the dAIEDGE-VLab, a collaborative virtual lab designed to facilitate artificial intelligence experimentation on edge devices.
This virtual environment allows researchers and data scientists to launch AI models on different specialised hardware platforms, without the need for direct physical access to the device.
dAIEDGE-VLab already has a first key functionality: a remote benchmarking service. Through this system, users can submit a pre-trained model via a graphical interface. The model is automatically run on different available platforms - from low-power controllers such as those from ST and NXP, to high-performance devices and dedicated AI accelerators - and the performance results are returned to the user.
This allows different implementations to be efficiently compared, metrics such as latency or throughput to be contrasted, and the best option to be decided based on the needs of the use case. All this from a single, remotely accessible platform.
Discover the full interview by clicking here or on our YouTube channel.