How to use the trained model on brand new data?

Part 1 (2019)
1

I’ve just trained a model to classify images and it’s working nicely. What I want to do now is use that model for brand new, unlabelled, images. Not in the train, test or val data sets but brand new images.

I can’t figure out how to do this - do I need to load them into an ImageDataBunch? Because that will then require splitting them up into test/train sets, which I don’t want to do. Ideally what I’d like to do is model.predict(filepath) and it just spits out its prediction for the image but that doesn’t seem possible?

2

Check out the “Putting your model into production” section of the lesson 2 notebook.

3

whoops don’t know how I missed that!! Worked perfectly, thank you :slight_smile:

1 Like