KubeEdge Blueprint provides a Machine Learning Inference Offload application. The architecture is illustrated in the following picture.
This application provides a service that offload inference on the Edge. It consists three layers cloud, edge and device collaboration.
- Cloud - training application, focusing on Emotion recognition model training.
- Edge - inference application, provideing a ML offloading service for inference.
- Device - capture and preprocess the image and use edge ML offloading service to provide an emotion recognition service.
KubeEdge landing application deployment guide
The following picture shows detailed workflow of the KubeEdge landing application:
