Q: Why inference results generated by AI Inference Server were not collected by Vision Data Collector successfully?
A: To guarantee inference results be collected by Vision Data Collector successfully the following requirements should be fulfilled:
If Zero Message Queue was used in AI Inference Server for inference result output, make sure the host’s name is “visiondatacollector-service” when configuring data connection in AI Inference Server and using it as “Connection” to publish inference results.
If Databus was used in AI Inference Server for inference result output, make sure that the username and password were correct when configure data connection in Vision Data Collector. And also make sure the authorization of Databus type connection is configured at Setup page of Vision Data Collector.
Make sure the output of pipeline package was correctly configured (the keys “image_timestamp” and “sensor_id” were included to the output of pipeline package according to correct structure as described in user manual).
Assuming your pipeline publishes inference result by variable name “result”, and this “result” is a Python dict type, the following picture shows how these two items are added.
Make sure the correct pipeline was selected in Vision Data Collector and this pipeline is running in AI Inference Server and generating inference results.
Parent topic:Frequently Asked Questions