1
$\begingroup$

I'm new to Caffe. I'm trying to understand the rules of using convolutional neural networks. And my questions are:

  • is it mandatory for the images to be small for training?
  • is it mandatory for the images to be squared?

I'm asking this because the biggest example of images used I found is 512 x 512. Would it be counter-intuitive to use for example HD size images?

$\endgroup$
  • 1
    $\begingroup$ NO for both questions. $\endgroup$ – SmallChess Feb 2 '17 at 10:32
  • $\begingroup$ Then why all examples use these settings ? $\endgroup$ – Amani Feb 2 '17 at 10:34
  • 1
    $\begingroup$ 1. Tutorial examples must be small for you to run. 2. There was no particular reason to go for anything but squares. $\endgroup$ – SmallChess Feb 2 '17 at 10:35
1
$\begingroup$

As pointed out in the comments, both are not necessary. When making an example it is howver very convenient to disregard aspects. Starting with the basics is the key to bring understanding. When you understand how it works it is easy to see that the size of the images are not an issue with regard to the network architecture. The image size does have an impact computationally however.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.