[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[ProgSoc] Using Neural Nets to process images



Hi, I have a question for all you neural net (and back prop.) enthusiast out there. I have a small project in mind; the project is using a neural net, input an image on the NN's input nodes (scalled accordingly) and the output would be the description of the image (this would also be scaled accordingly) (e.g. banana, orange, mouse).
Now some assumptions:
- The image has no background(or a black background) and contains only a single 300x400 2D image.
- The same image might be used several times in the NN but from different angles or with different brightness.
- All scalling of the input and output nodes have been taken care of so that they may fit into the NN.
The problem before i start is that has a project like this (or a similar one) been done before? and secondly at a glance what issues might occur? (aside from the usual overfitting and local minima)
Thanks
Stan


-
You are subscribed to the progsoc mailing list. To unsubscribe, send a
message containing "unsubscribe" to progsoc-request@xxxxxxxxxxxxxxxxxxx
If you are having trouble, ask owner-progsoc@xxxxxxxxxxxxxxxxxx for help.