More elaborate documentation for TFMA

[WoutRuyters]

System information

• Have I written custom code (as opposed to using a stock example script provided in TensorFlow Model Analysis): No
• OS Platform and Distribution (e.g., Linux Ubuntu 16.04): Linux-4.9.0-4-amd64-x86_64-with-Ubuntu-18.04-bionic
• TensorFlow Model Analysis installed from (source or binary): source
• TensorFlow Model Analysis version (use command below): 0.29.0
• Python version: 3.6.9
• Jupyter Notebook version: 6.2.0

Describe the problem

I have been trying to apply TensorFlow Model Analysis to a model of mine in which the dataset exists of images. I have found however that the documentation is (in my opinion) quite lacking in quite a few parts of tfma.x I have also found that most example notebooks use text data, I started by following along this example here. But the problem is that my image data doesn't get decoded before being evaluated, resulting in problems. Are there different tools I should be using when working with images when I want to evaluate my models to prevent this from happening?

So in general I would like a few more example notebooks using tfma and evaluating model fairness on an image dataset and would like to see the tfma documentation be expanded since this is lacking in quite a few parts.