Inquiry in details of training datacentric model

Dear Organizer,

I would like to ask about the settings you use in training the data-centric model. Is it that you performed 30 random transforms for each image for training as you did with the preprocess script in the github?

Yours sincerely, Lennon

Hi Lennon,

we performed 50 random augmentations for each sample (samples_per_file param here).

Best Jakob

Thank you very much!