How to predict and save the prediction into the kagge format

vad · May 9, 2017, 6:42pm

Dear all,

so I just started my journey throw the fast ai course.

And I have a simple question:

how to predict cats and dogs for the Kaggle challenge?
I mean how to apply my model to the test data and get a file which I can submit.

As far as I understood, I have to use:

vgg.predict, but I’m not sure…

Kind regards,
Vad

Jason · May 9, 2017, 8:41pm

Hi @vad congrats on starting the course! I’m near the end of my part 1 journey.

If you’ve finished lesson 1, and assuming you’re using the vgg16.py file provided, take some time to review it. Your answer is in there

Also, regarding the format, you basically have to create an output csv file to be similar to the sample submission file that kaggle provides (many ways to do this). You can also take a glance at the next lesson (dogs_cats_redux.ipynb) but try to figure it out yourself first and then look.

Hope that helps.
Jason

github.com

fastai/courses/blob/master/deeplearning1/nbs/vgg16.py

from __future__ import division, print_function

import os, json
from glob import glob
import numpy as np
from scipy import misc, ndimage
from scipy.ndimage.interpolation import zoom

from keras import backend as K
from keras.layers.normalization import BatchNormalization
from keras.utils.data_utils import get_file
from keras.models import Sequential
from keras.layers.core import Flatten, Dense, Dropout, Lambda
from keras.layers.convolutional import Convolution2D, MaxPooling2D, ZeroPadding2D
from keras.layers.pooling import GlobalAveragePooling2D
from keras.optimizers import SGD, RMSprop, Adam
from keras.preprocessing import image

# In case we are going to use the TensorFlow backend we need to explicitly set the Theano image ordering
from keras import backend as K

This file has been truncated. show original

vad · May 10, 2017, 7:58am

Hi Jason,

thank you for the hint

vad · May 11, 2017, 2:41pm

for those who will come here for answers, I recommend to check out keras.predict_generator https://keras.io/models/sequential/. Also, the dogs_cats_redux notebook is basically the answer.