Prodigy and Skin Cancer Detection

[KSK (Srini)]

Hello,
I am new to Prodigy. I would like to know if there is a way to use Prodigy for Skin Cancer detection with Deep Learning Models. Can you please provide any resources regarding it?
Thank you

[ryanwesslen]

hi @KSK!

Thanks for your message!

Out-of-the-box, Prodigy works with spaCy which focuses on NLP/information retrieval problems. However, Prodigy was designed to be extensible that developers can custom Prodigy for their workflows like with PyTorch or TensorFlow, which are popular libraries for image recognition.

First, I'd start with our computer vision docs where we have a section on how to develop a custom Prodigy recipe with a custom image model.

As an example, we have a tutorial on creating a custom TensorFlow object recognition recipe for Prodigy based on TensorFlow's object recognition API. You can view the code here.

Next, I'd also recommend this post from Ines on good tips of working with PyTorch and Prodigy's bounding box annotations:

Pytorch based bounding box annotations

Hi and sorry you were having trouble with TensorFlow! If you already have a PyTorch model, it's probably better to just use that (Also, the modelling it typically the "hard part" so if you have a model that's trained and that predicts something, it should hopefully be no problem to integrate it with Prodigy). For your annotation workflow, you probably want to use a custom recipe with the image_manual interface. Here's a basic template you can use to get started: Funda…

Also, I would recommend looking at Prodigy's docs on input data formatting for images.

I don't have a lot of experience with skin cancer detection models, but I've found some online PyTorch resources that fine-tune general purpose image models. This may be a good approach to start with.

Hope these help to get you started!

[KSK (Srini)]

Thank you very much, @ryanwesslen @Ines Montani