Problem in DL text-non text classification contest level 1 sample solution

I am able to understand the sigmoid class and other few fundamentals part of code. But I not really get the code
given in the code present in image below. Please, tell me what is exactly done in this code.

Image:

This is just some preprocessing being done. Can you please share which line are you facing doubt in?

In the below image I am not understanding what the ā€œLā€ is doing to our image

aa

I mean what exactly is it converting

convert() is a PIL Library function, here the argument "L" is used to convert the image to grayscale, as it has only 1 channel.