Multi-dimension Output

[LEE__JS1]

Hello,

I am suffering for dealing with the Multi-dimensional Output in Openvino.

As far as I found, the output data could be obtained with the code below.

>> const auto data = inferRequest.GetBlob(LprOutputName)->buffer().as<float*>();

But how can I deal with the multi-dementional output ?

For example, the output could be 1-1-2-3 or 1-2-3-4.

Can anybody help me ?

[JesusE_Intel]

Hi JS,

It depends on the model, for example face-detection-adas-004 from the Intel pre-trained models. The output is a blob with shape [1,1,N,7] each detection has the following format:

[image_id, label, conf, x_min, y_min, x_max, y_max], where:

image_id - ID of the image in the batch

label - predicted class ID

conf - confidence for the predicted class

(x_min, y_min) - coordinates of the top left bounding box corner

(x_max, y_max) - coordinates of the bottom right bounding box corner.

 https://github.com/opencv/open_model_zoo/blob/master/models/intel/face-detection-retail-0004/description/face-detection-retail-0004.md#outputs

Take a look at the Interactive Face Detection Demo source on how the output blob is handled.

Regards,

Jesus

[LEE__JS1]

Dear Jesus E. (Intel),

Sorry for late reply and so much thanks to you for your message.

I would like to download the model you mentioned and test with it.

After the test, I will let you know about the result again.

Best regards,

JS