Human Protein Atlas Competition starter code

[ste (Stefano Giomo)]

Hi @wdhorton

Great model!
I’ve tried to address the same competition, but even using transfer learning on Y channel too, the model seems to overfit without reaching your score…
Comparing the two models it seems that you’re using a model way more complex than resnet50: it seems to mix two copies of resnet50 one with first convolution with 3-channel images and one with 4-channels images as input. Why do you do that?

Here is my work:

Classifier with multiple images as input and multiple labels as output

Here is the updated model with the option to use transfer learning with multiple channel images (any number of channels). It duplicates RGB weights of first convolution on top of missing channels (IE: RGBY image use RGBR weights, RGBXYZ use RGBRGB). TODO: Data Block Api support…