Not understood the concept

#1

in 7:04 why we are again deviding pixel by 255?? isnt the image will become too small to identify??

#2

We are normalising our image data, basically, bringing it in the range 0 to 1 so as to reduce the variance in our data (recall the reasons why we normalise data). That’s what we are doing here. We are shrinking the entire 0 to 255 range if values to 0 to 1. So, now an image with (1,1,1) of RGB value would represent white.

I hope this clears your doubt.