Add data downloading

data/download_mnist.py

@@ -0,0 +1,60 @@
+import os
+import numpy as np
+import subprocess
+from scipy.io import loadmat, savemat
+from skimage.transform import resize
+
+def mnist_resize(x):
+ H, W, C = 32, 32, 3
+ x = x.reshape(-1, 28, 28)
+ resized_x = np.empty((len(x), H, W), dtype='float32')
+ for i, img in enumerate(x):
+ # resize returns [0, 1]
+ resized_x[i] = resize(img, (H, W), mode='reflect')
+
+ # Retile to make RGB
+ resized_x = resized_x.reshape(-1, H, W, 1)
+ resized_x = np.tile(resized_x, (1, 1, 1, C))
+ return resized_x
+
+def main():
+ if os.path.exists('mnist.npz'):
+ print "Using existing mnist.npz"
+
+ else:
+ print "Opening subprocess to download data from URL"
+ subprocess.check_output(
+ '''
+ wget https://s3.amazonaws.com/img-datasets/mnist.npz
+ ''',
+ shell=True)
+
+ if os.path.exists('mnist32_train.mat') and os.path.exists('mnist32_test.mat'):
+ print "Using existing mnist32_train.mat and mnist32_test.mat"
+
+ else:
+ print "Resizing mnist.npz to (32, 32, 3)"
+ data = np.load('mnist.npz')
+ trainx = data['x_train']
+ trainy = data['y_train']
+ trainx = mnist_resize(trainx)
+ savemat('mnist32_train.mat', {'X': trainx, 'y': trainy})
+
+ testx = data['x_test']
+ testy = data['y_test']
+ testx = mnist_resize(testx)
+ savemat('mnist32_test.mat', {'X': testx, 'y': testy})
+
+ print "Loading mnist32_train.mat for sanity check"
+ data = loadmat('mnist32_train.mat')
+ print data['X'].shape, data['X'].min() ,data['X'].max()
+ print data['y'].shape, data['y'].min() ,data['y'].max()
+
+ print "Loading mnist32_test.mat for sanity check"
+ data = loadmat('mnist32_test.mat')
+ print data['X'].shape, data['X'].min() ,data['X'].max()
+ print data['y'].shape, data['y'].min() ,data['y'].max()
+
+
+if __name__ == '__main__':
+ main()

data/download_svhn.py

@@ -0,0 +1,31 @@
+import subprocess
+import os
+from scipy.io import loadmat 
+
+def main():
+ if os.path.exists('test_32x32.mat') and os.path.exists('train_32x32.mat'):
+ print "Using existing data"
+
+ else:
+ print "Opening subprocess to download data from URL"
+ subprocess.check_output(
+ '''
+ wget https://ufldl.stanford.edu/housenumbers/train_32x32.mat
+ wget https://ufldl.stanford.edu/housenumbers/test_32x32.mat
+ wget https://ufldl.stanford.edu/housenumbers/extra_32x32.mat
+ ''',
+ shell=True)
+
+ print "Loading train_32x32.mat for sanity check"
+ data = loadmat('train_32x32.mat')
+ print data['X'].shape, data['X'].min() ,data['X'].max()
+ print data['y'].shape, data['y'].min() ,data['y'].max()
+
+ print "Loading test_32x32.mat for sanity check"
+ data = loadmat('test_32x32.mat')
+ print data['X'].shape, data['X'].min() ,data['X'].max()
+ print data['y'].shape, data['y'].min() ,data['y'].max()
+
+
+if __name__ == '__main__':
+ main()