[Theano] Trying to classify Apples and Bananas with CNN ? Algorithm wont run

Hi Theano forum! Im having a problem with the Theano tutorial about Convolutional neural network with MNIST.

I changed the code so i would be able to test with my own images.

So here's what i did :

I got 10 apple images and resized them to 100x100 I got 10 banana Imagesand resized them to 100x100

I managed to create a CSV file. But im not sure i did it correctly. On the first line i did : |Class| | | Then i did : |0| |apple_0.bmp| and so on...

I packed them on pickle.

Then i changed the code parameters to support my images.

Now when i try to run, this happens :

C:SciSoftWinPython-64bit-2.7.9.4python-2.7.9.amd64python.exe A:/Code/MeusProjetosPython/TheanoTest/ConvolutionSum/convolutional_mlp.py ... loading data ... building the model ... training Optimization complete. Best validation score of inf % obtained at iteration 1, with test performance 0.000000 % The code for file convolutional_mlp.py ran for 0.00m

Process finished with exit code 0

I pushed the project here : BitBucket Project

https://bitbucket.org/andrelopes1705/theanoconv/src/e481d2f3b331?at=default

Would anyone look at it for me please! Any tip is appreciated!

convolutional_mlp.py to run the project pack.py to pack the data file.pkl.gz is the data that i packed images folder is the images im using and it also contains the CSV : CSV CSV : https://bitbucket.org/andrelopes1705/theanoconv/src/e481d2f3b3315269c3e128dde3254f7a92b77755/Images/labels.csv?at=default