#papayaprog

I guess you'd have to ask scikit-learn? I've never used it myself (I'm used to doing everything with TensorFlow and my own code). There might be a simple answer. I think that scikit-learn and TensorFlow can do about the same things.

EDIT: OK, I looked it up, and here is a bit from their website:

Let's take the example of face recognition. We would like to measure the effectiveness of a particular method of face recognition. Here is the TF Python API for loading images: tf.load("MNIST_images.txt"). The code above loads a text file called "MNIST_images.txt" that contains 60,000 low-resolution images of pencils, houses, faces, etc. The purpose of the "tf.load" function in TensorFlow is simply to load files into memory for us to access as tensors.

We can use a function called save_image to save an image as a file. Here is a Python call to save_image saving a MNIST image with label 2 to an image called "2.png."

I can't really understand any of this stuff, I'm just going by what it says on their website

I think what they are trying to say is "if you have a file of images, you can put it into tensorflow, load it up and work with it. If you want to work with the data as a single tensor, or do some matrix operations on the data, you should be able to do it with tensorflow."