Hopefully soon they will make a model that merges something like S3 and Grove Vision AI accelerator.

Tried running yolo pose on bare S3, and the results were underwhelming.