Most of the AI/ML learning and model development is being done in the cloud, typically on high-end x86 servers with powerful GPU cards. However, deploying these models to embedded Edge devices which have a much smaller footprint in terms of compute and memory can be a challenge. This session describes how the Edge differs from the cloud in terms of AI/ML processing. It goes on to describe a set of services that can work with different cloud AI/ML frameworks like AWS Sagemaker and Google AutoML, take their output model, convert it into a size and format suitable for embedded Edge devices and deploy over the air.
