How to generate images in same Dockerfile as classification model

[Adel Hassan (Adel)]

I'm a bit confused as to how to structure my project when preparing for submission. The final submission should have a single Dockerfile that generates images from waveforms and then also classifies them. However, the code to generate the EKG images is in a separate GitHub repository from the example code.

For training, I simply have both repositories on my computer and I run the code separately. How are y'all putting everything in one place? Are you just copying the EKG generator repo into the folder with the classification code (e.g., run_model.py, train_model.py, etc.)? Or is there a better solution that I haven't thought of?

[PhysioNet Challenge]

Hello,

Yes, you can copy the ECG image generation code from ECG-Image-Kit into the same folder with the rest of your entry and import and run it as part of your training code. The Dockerfile will install the dependencies for ECG-Image-Kit, although you may need to make a couple of minor changes depending on the base image for your entry.

There are other ways to do it, but this may be one of the simpler solutions.  We may want or need to change how we support this for the official phase, so feedback about what works and what doesn't work is very helpful.

Best,
Matt

(On behalf of the Challenge team.)

Please post questions and comments in the forum. However, if your question reveals information about your entry, then please email info at physionetchallenge.org. We may post parts of our reply publicly if we feel that all Challengers should benefit from it. We will not answer emails about the Challenge to any other address. This email is maintained by a group. Please do not email us individually.