Combining image + text for FastAI2

Hi guys,

I have some difficulties in combing multiple types of dataset; images + text.

As I know, in FastAI1, I can simply use MixedItemList for merging them together, but after I switched to FastAI2 I can’t find any tutorials for doing this. Also, I take a look at this source It works only tabular + images.

Does anyone know which class or function in FastAI2 is equivalent to MixedItemList ?

Cheers!

That approach works with anything. Hence the (and any other type) I’ve used it to combine text and tabular before with no issue. Text and image should be simplistic as well.

Hi, @muellerzr

When we call show_batch(), are those tabular + image pair must be depended on each other?

For example; if we call show_batch size of 2.

  1. tabA, targetA
  2. tabB, targetB
  3. imgA, targetA,
  4. imgB, targetB

Would row 1 and 3, as well as 2 and 4 be referring to each other?
or it’s just random among them