In this document, we will give a guide about deploying an application step by step in OTE-stack platform on AI Edge for testing.
3 Arm servers for IEC edge infrastructure and another server which have installed harbor are needs. A face recognition application, which can be found at https://hub.docker.com/r/otestack/face-detect-demo, is used to validate the functions of the OTE. Before start testing, you need to push the image to the harbor server.
For test architecture, We use one IEC edge cluster with 3 nodes as the basic infrastructure and install the OTE-stack platform in it. After everything is ready, we test the function of deploy a application through the OTE website which is installed in the master node of IEC cluster.
In this test scenarios, a face recognition application will be deploy to the IEC cluster via browser. The app is a server that detects the face in an image sent by client and outputs the position of objects. The usage of the server is shown in Step 4.
Send request to the face server through the ip:port got by last step
$ # prepare image $ wget http://aip.bdstatic.com/portal-pc-node/dist/1590550949362/images/technology/face/detect/demo-card-2.jpg $ # make a request $ image=demo-card-2.jpg $ echo '{"image": "'$(base64 -w 0 $image)'"}' > data.json $ curl -X POST 10.247.22.115:8080/face_detect -d@data.json {"objects": [{"location": {"x1": 898, "y1": 217, "x2": 1154, "y2": 518}, "prob": 0.9999696016311646}, {"location": {"x1": 444, "y1": 331, "x2": 700, "y2": 657}, "prob": 0.9997757077217102}]} |