I want to create a convolutional neural network to identify a particular object in an image, say a guitar. I know there are a lot of posts and answers about this but I'm very new to this field and am not being able to find a simple explanation. If I have the coordinates for any guitar in an image how do I convert that to vector/tensor can someone please tell me?? Just a few steps as to what I should do next would be very helpful! Thanks!
$\begingroup$
$\endgroup$
1
-
$\begingroup$ you need a lot of guitar images. there's a large number of examples in internet with MNIST sample, that recognizes digits. I'd start with them $\endgroup$– AksakalCommented Oct 31, 2017 at 13:15
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
To train a neural network you need lots of examples (10'000s typically) where you have the correct answer.
More generally, I like the explanation by Ujjwal Karn and Adit Deshpande.
After that, Tensorflow has good tutorials. Starting with this will be a good start so that you will understand the following tutorials on CNN's (this and this tutorial) better.
