In his approach, deploying lightweight models and using edge computing or distributed inference nodes helps reduce latency. Especially in mobile …
http://dlvr.it/TMPh85