How to systematically determine necessary steps for input and ouput preparation?

[Lange__Benjamin]

Hello OpenVino Community,

I am trying to code a minimal SSD detection example on my own using OpenVino in the current Version 2020.4.

I have read the following resources:
https://docs.openvinotoolkit.org/latest/openvino_docs_IE_DG_inference_engine_intro.html
https://docs.openvinotoolkit.org/latest/openvino_docs_IE_DG_Integrate_with_customer_application_new_API.html

In the latter resource in step 3 you have to prepare inputs and outputs. I could not find a more detailed resource on this matter.

When I look at this Object Detection SSD example the steps 6 and 9 are about input preparation, but how do I know what steps I have to take? Are the presented steps always necessary? I am somehow missing the link between the IR topology and actually coding (not copy pasting) the input and output preparation in a well understood manner.

Any approaches are welcome! As well as pointers to more detailed resources.

[Munesh_Intel]

Hi Bernard,

I would suggest you view the following videos, which will provide your more insights into Inference Engine workflow.

The Inference-Engine Concept | OpenVINO™ toolkit | Ep. 15 | Intel Software

https://www.youtube.com/watch?v=e6R13V8nbak&list=PLDKCjIU5YH6jMzcTV5_cxX9aPHsborbXQ&index=15

Inference-Engine API Sample Code | OpenVINO™ toolkit | Ep. 16 | Intel Software

https://www.youtube.com/watch?v=JwgPgKClU3w&list=PLDKCjIU5YH6jMzcTV5_cxX9aPHsborbXQ&index=16

Inference Engine Python Sample | OpenVINO™ toolkit | Ep. 17 | Intel Software

https://www.youtube.com/watch?v=6Dzvamu3mg8&list=PLDKCjIU5YH6jMzcTV5_cxX9aPHsborbXQ&index=17

Regards,

Munesh

[Munesh_Intel]

Hi Benjamin,

This thread will no longer be monitored since we have provided links to resources. If you need any additional information from Intel, please submit a new question.

Regards,

Munesh