Click here to Skip to main content
15,945,913 members
Please Sign up or sign in to vote.
1.00/5 (1 vote)
See more:
Im working on an ANN project in which iam asked to classify and train a network that has 3 images as an input.
These images are: apple, banana, and a mobile.
The images has to be inputed pixel by pixel, and each image be trained in a way that the system must know that a red apple is an apple, or the green apple is an apple, etccc.

Second, I dont know how many hidden layers and how many neurons in each hidden layer I have to take (I started with one hidden layer and 6 hidden neurons.

It is a must that we use the Back propagation algorithm for the classification of the three images.

I Am writing the code in C#, but I have some problems in coding the project especially how to tell the system that it must know that the green and the yellow apples are apples and will not be able to recognize the green apple since we trained the system to recognize the red apple. ( I want it to recognize all colored apples as apples).

Also, assuming that Iam taking a 256 * 256 image, and each pixel is an input, then we have around 65000 input, which means we have 65000 weights for each neuron in the hidden layer, and I started with 6 neurons. Does it make sense?

Could any one help in the coding please?
Updated 21-May-11 2:16am

1 solution

It seems to me that an apple is round, a banana bent, and a mobile is square. Beyond that, if it's your school project, I don't think it's appropriate for people to give you code, unless it's in specific reply to specific code that you post here.
Share this answer
Mohammad Shehab 21-May-11 8:18am    
so if an apple is round, a banana bent, and a mobile is square, what should i do?
Mohammad Shehab 22-May-11 15:50pm    
Ok, I think that this idee gave the solution to my problem. Thanks!

This content, along with any associated source code and files, is licensed under The Code Project Open License (CPOL)

CodeProject, 20 Bay Street, 11th Floor Toronto, Ontario, Canada M5J 2N8 +1 (416) 849-8900