Bowl kaggle competition (More info on kaggle website http://www.kaggle.com/c/datasciencebowl) Predict ocean health, one plankton at a time
##Software: Language:python This code prepare the images for the classification problem, do training usign CNN with nolearn, theano, lasagne. Credits to the posts on D. Nouri blog http://danielnouri.org/notes/
There are 3 files: ###Prepare_Features.py: Useful funtions for image preprocessing. The results of preprocessing and resizing are saved on disk as np.array with labels from 0 to 120. The encoded labels printed can be used for submission. ###Training-NET.py Build a Neural Net with different layers, based on Nouri setup, train the net and save the model on disk ###Prediction-NET.py Load the saved model and perform prediction, writing them in a file for submission.
You can use a json file to set the PATH and other parameters once for all.
####Other useful repository links https://github.com/msegala/Kaggle-National_Data_Science_Bowl