ImageClassifier Predict Label

[aribornstein]

🚀 Feature

Prediction Currently Returns the class index based on the folder position in train. It would be great to be able to able to see the targets.

Motivation

It's hard to interpret class id numbers when you have more than a few classes.
model.predict(["cat.png"], target=true)
output -> 1

What does one mean?

Pitch

model.predict(["cat.png"], target=true)
output ->cat

Alternatives

Open to discussion

Additional context

[ibraheemmmoosa]

Another alternative can be to have an easy API to convert between ID and class label. The ID is important if you want your output to be numerical.

[aribornstein]

Adding to this,a way to get the prediction class distribution.

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[ethanwharris]

Fixed by #229 - can now use model.serializer = Labels() or model.serializer = Probabilities(), for other options see the predictions docs.