Run cascaded DLK package in Mech Vision

Q&A
1

I have developed a cascaded DLK model using Instance Segmentation - Classification function. May I know how to display the classification result instead of the instance segmentation result?

Please find the following image for your reference. Even though I export the DLK in the insSeg_objDet model, but I can still only get instance segmentation results in Mech Vision.

Capture
Capture1408×473 24.8 KB

2

When you inport a cascaded DLK model in Mech-Vision, the Deep learning Model Package Inference should be like this:

image
image1134×558 35.8 KB

You can see two different output port “DLValue/InstanceSegmentation” and “DLValue/Classfication”

After that you need to add a step “Deep Learning Result Parser”:

image
image745×328 12.4 KB

Select the type you want to parse in the step parameter

The process flow can be reference as:

image
image998×610 33.3 KB

3

Thanks for your explanation. For the deep learning model that I exported from Mech-DLK, I could not see the different output ports like yours when I use in Mech Vision. Is there any extra settings that need to be done to export a cascaded deep learning model?

4

If your Mech-Vision version is too low, it could lead to this issue. In case you are using the latest version （1.7.4) ， the problem might stem from an incorrectly trained deep learning model package