Validate user-traced letters in canvas with Personal Image Classifier

I am brand new to MIT App Inventor (no real programming background/self taught) and also trying to teach myself Korean. I was inspired to create an app to display a hangul consonant vowel pair, let the user trace it, and then let the user know if they traced it correctly.

I made images with a light grey text of each of the hangul letter pairs and uploaded them as media. I used canvas to display the image as a background and was successful in coding tracing the letters.

I was able to make training data and uploaded the mdl file for the personal image classifier .

Where I am stuck is saving the canvas image and then classifying the saved image. I’m not sure how to properly use the blocks for this, or if I am using the right tools.

Can canvas user drawings be saved directly to where the Personal Image Classifier can use it? If a canvas drawing is saved to device storage, can it be deleted after it is classified? I see how the File component allows for deletions, but does canvas save to where the file component can delete it?

Any assistance would be much appreciated. All examples I have tried to look at use the camera for classification and not canvas.

Much thanks!

On my android 10 phone this is working:

See MIT’s comments here Saving photos in the Personal Image Classifier - #2 by ewpatton