def execute(outputsize):
faces_dir_path = "data/train_set/48_48_faces_web_augmented"
bkgs_dir_path = "data/train_set/48_48_nonfaces_aflw"
target_path = "data/train_set/13"
faces_dir=join(target_path,"faces")
nonfaces_dir = join(target_path,"nonfaces")
os.makedirs(nonfaces_dir)
os.makedirs(faces_dir)
img_faces = [ f for f in listdir(faces_dir_path) if isfile(join(faces_dir_path,f)) and f.endswith("png") ]
img_bkgs = [ f for f in listdir(bkgs_dir_path) if isfile(join(bkgs_dir_path,f)) and f.endswith("jpg") ]
for i, img_name in enumerate(img_faces):
img_path = join(faces_dir_path,img_name)
img = imread(img_path)
resized_img = resize(img,outputsize)
ubyte_img = img_as_ubyte(resized_img)
imsave(join(faces_dir,img_name), ubyte_img)
print "processed "+ img_path
for i, img_name in enumerate(img_bkgs):
img_path = join(bkgs_dir_path,img_name)
img = imread(img_path)
gray_img = rgb2gray(img)
resized_img = resize(gray_img,outputsize)
ubyte_img = img_as_ubyte(resized_img)
imsave(join(nonfaces_dir,img_name), ubyte_img)
print "processed "+ img_path
def test_bool_and_anti_aliasing_errors():
img = np.zeros((10, 10), dtype=bool)
with pytest.raises(ValueError):
rescale(img, 0.5, anti_aliasing=True)
with pytest.raises(ValueError):
resize(img, (5, 5), anti_aliasing=True)
def test_bool_nonzero_order_errors(order):
img = np.zeros((10, 10), dtype=bool)
with pytest.raises(ValueError):
rescale(img, 0.5, order=order)
with pytest.raises(ValueError):
resize(img, (5, 5), order=order)
with pytest.raises(ValueError):
warp(img, np.eye(3), order=order)
def test_resize3d_keep():
# keep 3rd dimension
x = np.zeros((5, 5, 3), dtype=np.float64)
x[1, 1, :] = 1
resized = resize(x, (10, 10), order=0, anti_aliasing=False,
mode='constant')
with pytest.raises(ValueError):
# output_shape too short
resize(x, (10, ), order=0, anti_aliasing=False, mode='constant')
ref = np.zeros((10, 10, 3))
ref[2:4, 2:4, :] = 1
assert_array_almost_equal(resized, ref)
resized = resize(x, (10, 10, 3), order=0, anti_aliasing=False,
mode='constant')
assert_array_almost_equal(resized, ref)
def test_resize2d():
x = np.zeros((5, 5), dtype=np.float64)
x[1, 1] = 1
resized = resize(x, (10, 10), order=0, anti_aliasing=False,
mode='constant')
ref = np.zeros((10, 10))
ref[2:4, 2:4] = 1
assert_array_almost_equal(resized, ref)
def test_downsize():
x = np.zeros((10, 10), dtype=np.double)
x[2:4, 2:4] = 1
scaled = resize(x, (5, 5), order=0, anti_aliasing=False, mode='constant')
assert_equal(scaled.shape, (5, 5))
assert_equal(scaled[1, 1], 1)
assert_equal(scaled[2:, :].sum(), 0)
assert_equal(scaled[:, 2:].sum(), 0)
def test_bool_array_warnings():
img = np.zeros((10, 10), dtype=bool)
with expected_warnings(['Input image dtype is bool']):
rescale(img, 0.5, anti_aliasing=True)
with expected_warnings(['Input image dtype is bool']):
resize(img, (5, 5), anti_aliasing=True)
with expected_warnings(['Input image dtype is bool']):
rescale(img, 0.5, order=1)
with expected_warnings(['Input image dtype is bool']):
resize(img, (5, 5), order=1)
with expected_warnings(['Input image dtype is bool']):
warp(img, np.eye(3), order=1)
def test_bool_img_resize():
img = np.ones((12, 18), dtype=bool)
img[2:-2, 4:-4] = False
res = resize(img, (6, 9))
expected = np.ones((6, 9))
expected[1:-1, 2:-2] = False
assert_array_equal(res, expected)
def test_resize3d_resize():
# resize 3rd dimension
x = np.zeros((5, 5, 3), dtype=np.float64)
x[1, 1, :] = 1
resized = resize(x, (10, 10, 1), order=0, anti_aliasing=False,
mode='constant')
ref = np.zeros((10, 10, 1))
ref[2:4, 2:4] = 1
assert_array_almost_equal(resized, ref)
def test_order_0_warp_dtype(dtype):
img = _convert(astronaut()[:10, :10, 0], dtype)
assert resize(img, (12, 12), order=0).dtype == dtype
assert rescale(img, 0.5, order=0).dtype == dtype
assert rotate(img, 45, order=0).dtype == dtype
assert warp_polar(img, order=0).dtype == dtype
assert swirl(img, order=0).dtype == dtype
def test_resize3d_2din_3dout():
# 3D output with 2D input
x = np.zeros((5, 5), dtype=np.float64)
x[1, 1] = 1
resized = resize(x, (10, 10, 1), order=0, anti_aliasing=False,
mode='constant')
ref = np.zeros((10, 10, 1))
ref[2:4, 2:4] = 1
assert_array_almost_equal(resized, ref)
def test_resize_nd():
for dim in range(1, 6):
shape = 2 + np.arange(dim) * 2
x = np.ones(shape)
out_shape = np.asarray(shape) * 1.5
resized = resize(x, out_shape, order=0, mode='reflect',
anti_aliasing=False)
expected_shape = 1.5 * shape
assert_array_equal(resized.shape, expected_shape)
assert np.all(resized == 1)
def test_resize2d_4d():
# resize with extra output dimensions
x = np.zeros((5, 5), dtype=np.float64)
x[1, 1] = 1
out_shape = (10, 10, 1, 1)
resized = resize(x, out_shape, order=0, anti_aliasing=False,
mode='constant')
ref = np.zeros(out_shape)
ref[2:4, 2:4, ...] = 1
assert_array_almost_equal(resized, ref)
def test_downsize(dtype):
x = np.zeros((10, 10), dtype=dtype)
x[2:4, 2:4] = 1
scaled = resize(x, (5, 5), order=0, anti_aliasing=False, mode='constant')
expected_dtype = np.float32 if dtype == np.float16 else dtype
assert scaled.dtype == expected_dtype
assert scaled.shape == (5, 5)
assert scaled[1, 1] == 1
assert scaled[2:, :].sum() == 0
assert scaled[:, 2:].sum() == 0
def test_nonzero_order_warp_dtype(dtype, order):
img = _convert(astronaut()[:10, :10, 0], dtype)
float_dtype = _supported_float_type(dtype)
assert resize(img, (12, 12), order=order).dtype == float_dtype
assert rescale(img, 0.5, order=order).dtype == float_dtype
assert rotate(img, 45, order=order).dtype == float_dtype
assert warp_polar(img, order=order).dtype == float_dtype
assert swirl(img, order=order).dtype == float_dtype
def test_resize_dtype():
x = np.zeros((5, 5))
x_f32 = x.astype(np.float32)
x_u8 = x.astype(np.uint8)
x_b = x.astype(bool)
assert resize(x, (10, 10), preserve_range=False).dtype == x.dtype
assert resize(x, (10, 10), preserve_range=True).dtype == x.dtype
assert resize(x_u8, (10, 10), preserve_range=False).dtype == np.double
assert resize(x_u8, (10, 10), preserve_range=True).dtype == np.double
assert resize(x_b, (10, 10), preserve_range=False).dtype == bool
assert resize(x_b, (10, 10), preserve_range=True).dtype == bool
assert resize(x_f32, (10, 10), preserve_range=False).dtype == x_f32.dtype
assert resize(x_f32, (10, 10), preserve_range=True).dtype == x_f32.dtype
def test_resize3d_bilinear():
# bilinear 3rd dimension
x = np.zeros((5, 5, 2), dtype=np.float64)
x[1, 1, 0] = 0
x[1, 1, 1] = 1
resized = resize(x, (10, 10, 1), order=1, mode='constant',
anti_aliasing=False)
ref = np.zeros((10, 10, 1))
ref[1:5, 1:5, :] = 0.03125
ref[1:5, 2:4, :] = 0.09375
ref[2:4, 1:5, :] = 0.09375
ref[2:4, 2:4, :] = 0.28125
assert_array_almost_equal(resized, ref)
def test_resize_clip(order, preserve_range, anti_aliasing, dtype):
# test if clip as expected
if dtype == np.uint8 and (preserve_range or order == 0):
expected_max = 255
else:
expected_max = 1.0
x = np.ones((5, 5), dtype=dtype)
if dtype == np.uint8:
x *= 255
resized = resize(x, (3, 3), order=order, preserve_range=preserve_range,
anti_aliasing=anti_aliasing)
assert resized.max() == expected_max
def execute(outputsize):
faces_dir_path = "data/train_set/48_48_faces_web_augmented"
bkgs_dir_path = "data/train_set/48_48_nonfaces_aflw"
target_path = "data/train_set/13"
faces_dir = join(target_path, "faces")
nonfaces_dir = join(target_path, "nonfaces")
os.makedirs(nonfaces_dir)
os.makedirs(faces_dir)
img_faces = [
f for f in listdir(faces_dir_path)
if isfile(join(faces_dir_path, f)) and f.endswith("png")
]
img_bkgs = [
f for f in listdir(bkgs_dir_path)
if isfile(join(bkgs_dir_path, f)) and f.endswith("jpg")
]
for i, img_name in enumerate(img_faces):
img_path = join(faces_dir_path, img_name)
img = imread(img_path)
resized_img = resize(img, outputsize)
ubyte_img = img_as_ubyte(resized_img)
imsave(join(faces_dir, img_name), ubyte_img)
print "processed " + img_path
for i, img_name in enumerate(img_bkgs):
img_path = join(bkgs_dir_path, img_name)
img = imread(img_path)
gray_img = rgb2gray(img)
resized_img = resize(gray_img, outputsize)
ubyte_img = img_as_ubyte(resized_img)
imsave(join(nonfaces_dir, img_name), ubyte_img)
print "processed " + img_path
def test_downsize_anti_aliasing_invalid_stddev():
x = np.zeros((10, 10), dtype=np.float64)
with pytest.raises(ValueError):
resize(x, (5, 5), order=0, anti_aliasing=True, anti_aliasing_sigma=-1,
mode='constant')
with expected_warnings(["Anti-aliasing standard deviation greater"]):
resize(x, (5, 15), order=0, anti_aliasing=True,
anti_aliasing_sigma=(1, 1), mode="reflect")
resize(x, (5, 15), order=0, anti_aliasing=True,
anti_aliasing_sigma=(0, 1), mode="reflect")
def resize_imgs_in_dir(outputsize):
img_dir_path = "data/newnonfaces/48_48_non_faces_aflw"
target_path = "data/newnonfaces/13/nonfaces"
os.makedirs(target_path)
img_faces = [ f for f in listdir(img_dir_path) if isfile(join(img_dir_path,f)) and f.endswith("jpg") ]
for i, img_name in enumerate(img_faces):
img_path = join(img_dir_path,img_name)
img = imread(img_path)
resized_img = resize(img,outputsize)
ubyte_img = img_as_ubyte(resized_img)
imsave(join(target_path,img_name), ubyte_img)
print "processed "+ img_path
def resize_imgs_in_dir(outputsize):
img_dir_path = "data/newnonfaces/48_48_non_faces_aflw"
target_path = "data/newnonfaces/13/nonfaces"
os.makedirs(target_path)
img_faces = [
f for f in listdir(img_dir_path)
if isfile(join(img_dir_path, f)) and f.endswith("jpg")
]
for i, img_name in enumerate(img_faces):
img_path = join(img_dir_path, img_name)
img = imread(img_path)
resized_img = resize(img, outputsize)
ubyte_img = img_as_ubyte(resized_img)
imsave(join(target_path, img_name), ubyte_img)
print "processed " + img_path
def check_for_image():
import skimage
import skimage.data
import skimage.util
from skimage.color import rgb2gray
img = rgb2gray(imread("data/originalPics/2002/07/19/big/img_372.jpg"))
#img = rgb2gray(skimage.data.lena())
#img = imread("data/processed_images/13_train_set_aflw/train/faces/001111.jpg")
# Load mean value from file.
f = file(LOAD_MEAN_TO_FILE, 'rb')
mean_val = pickle.load(f)
f.close()
# Scale search
for mul in xrange(3, 20):
img_orig = resize(img, (mul * 10, mul * 10))
im = img_orig[..., np.newaxis]
arr = skimage.util.view_as_windows(im, (28, 28, 1), step=1)
arr = arr - mean_val
f = file(LOAD_STATE_FROM_FILE, 'rb')
obj = pickle.load(f)
f.close()
arr = np.rollaxis(arr, 5, 3)
borrow = True
shared_x = theano.shared(np.asarray(arr, dtype=theano.config.floatX),
borrow=borrow)
iT = T.lscalar()
jT = T.lscalar()
x = T.tensor3("x")
layer0_input = x.reshape((1, 1, 28, 28))
net = twelve_net(layer0_input, None, T.tanh, obj)
prediction = net.log_regression_layer.y_pred
py_x = net.log_regression_layer.p_y_given_x
test_model = theano.function([iT, jT],
[prediction, py_x, layer0_input],
givens={x: shared_x[iT, jT, 0, :, :, :]})
rows = arr.shape[0]
cols = arr.shape[1]
count = 0
faces = []
for i in xrange(rows):
for j in xrange(cols):
[y, p_y_given_x, f] = test_model(i, j)
f = f.reshape(1, 28, 28)
f = np.rollaxis(f, 0, 3)
#plt.imshow(f)
#plt.show()
if y == 1:
count += 1
faces.append([i, j])
print i, j
print("Check")
plt.imshow(img_orig)
plt.set_cmap('gray')
img_desc = plt.gca()
for point in faces:
topleftx = point[1]
toplefty = point[0]
rect = patches.Rectangle((topleftx, toplefty),
28,
28,
fill=False,
color='c')
img_desc.add_patch(rect)
print count
plt.show()
def check_for_image():
import skimage
import skimage.data
import skimage.util
from skimage.color import rgb2gray
img = rgb2gray(imread("data/originalPics/2002/07/19/big/img_372.jpg"))
#img = rgb2gray(skimage.data.lena())
#img = imread("data/processed_images/13_train_set_aflw/train/faces/001111.jpg")
# Load mean value from file.
f = file(LOAD_MEAN_TO_FILE, 'rb')
mean_val = pickle.load(f)
f.close()
# Scale search
for mul in xrange(3, 20):
img_orig = resize(img, (mul*10, mul*10))
im = img_orig[..., np.newaxis]
arr = skimage.util.view_as_windows(im, (28, 28, 1), step=1)
arr = arr - mean_val
f = file(LOAD_STATE_FROM_FILE, 'rb')
obj = pickle.load(f)
f.close()
arr = np.rollaxis(arr, 5, 3)
borrow = True
shared_x = theano.shared(np.asarray(arr, dtype=theano.config.floatX),
borrow=borrow)
iT = T.lscalar()
jT = T.lscalar()
x = T.tensor3("x")
layer0_input = x.reshape((1, 1, 28, 28))
net = twelve_net(layer0_input, None, T.tanh, obj)
prediction = net.log_regression_layer.y_pred
py_x = net.log_regression_layer.p_y_given_x
test_model = theano.function(
[iT, jT],
[prediction, py_x, layer0_input],
givens={
x: shared_x[iT, jT, 0, :, :, :]
}
)
rows = arr.shape[0]
cols = arr.shape[1]
count = 0
faces = []
for i in xrange(rows):
for j in xrange(cols):
[y,p_y_given_x,f] = test_model(i, j)
f=f.reshape(1, 28, 28)
f = np.rollaxis(f, 0, 3)
#plt.imshow(f)
#plt.show()
if y == 1:
count += 1
faces.append([i, j])
print i, j
print ("Check")
plt.imshow(img_orig)
plt.set_cmap('gray')
img_desc = plt.gca()
for point in faces:
topleftx = point[1]
toplefty = point[0]
rect = patches.Rectangle(
(topleftx, toplefty),
28,
28,
fill=False,
color='c'
)
img_desc.add_patch(rect)
print count
plt.show()
def check_for_image():
import skimage
import skimage.data
import skimage.util
img = skimage.data.lena()
#img = io.imread("data/originalPics/2002/07/19/big/img_130.jpg")
#img = io.imread("data/train_set/dataset/13/train/faces/00027998.png")
img = rgb2gray(img)
img = img_as_ubyte(img)
img = img[:, :, np.newaxis]
#img = imread("data/processed_images/13_train_set_aflw/train/faces/001111.jpg")
for mul in xrange(3, 20):
im = resize(img, (mul * 10, mul * 10))
im = img_as_ubyte(im)
arr = skimage.util.view_as_windows(im, (13, 13, NUM_CHANNELS),
step=STEP_SIZE)
f = file(LOAD_STATE_FROM_FILE, 'rb')
obj = pickle.load(f)
f.close()
arr = np.rollaxis(arr, 5, 3)
borrow = True
shared_x = theano.shared(
np.asarray(arr, dtype=theano.config.floatX), # @UndefinedVariable
borrow=borrow)
iT = T.lscalar()
jT = T.lscalar()
x = T.tensor3("x")
layer0_input = x.reshape((1, NUM_CHANNELS, 13, 13))
net = twelve_net(layer0_input, None, relu, obj)
prediction = net.log_regression_layer.y_pred
py_x = net.log_regression_layer.p_y_given_x
test_model = theano.function([iT, jT],
[prediction, py_x, layer0_input],
givens={x: shared_x[iT, jT, 0, :, :, :]})
rows = arr.shape[0]
cols = arr.shape[1]
count = 0
faces = []
fig, axarr = plt.subplots(1, 2)
confidence_map = np.zeros((rows, cols))
for i in xrange(rows):
for j in xrange(cols):
[y, p_y_given_x, f] = test_model(i, j)
f = f.reshape(NUM_CHANNELS, 13, 13)
f = np.rollaxis(f, 0, 3)
f = f[:, :, 0]
confidence_map[i, j] = p_y_given_x[0, 1]
# plt.imshow(f,cmap = "Greys_r")
# plt.show()
if y == 1:
count += 1
faces.append([i * STEP_SIZE, j * STEP_SIZE])
print i, j
print("Check")
# hack for gray
im = im[:, :, 0]
confidence_map = np.pad(confidence_map, 6, 'constant')
axarr[0].get_xaxis().set_visible(False)
axarr[0].get_yaxis().set_visible(False)
axarr[1].get_xaxis().set_visible(False)
axarr[1].get_yaxis().set_visible(False)
axarr[1].imshow(confidence_map, cmap="Greys_r")
axarr[0].imshow(im, cmap="Greys_r")
img_desc = plt.gca()
#plt.imshow(im,cmap = "Greys_r")
#img_desc = plt.gca()
for point in faces:
topleftx = point[1]
toplefty = point[0]
rect = patches.Rectangle((topleftx, toplefty),
13,
13,
fill=False,
color='c')
axarr[0].add_patch(rect)
print count
fig_name = "lena_" + str(mul) + ".png"
plt.savefig(fig_name, bbox_inches='tight')
def check_for_image():
import skimage
import skimage.data
import skimage.util
img = skimage.data.lena()
#img = io.imread("data/originalPics/2002/07/19/big/img_130.jpg")
#img = io.imread("data/train_set/dataset/13/train/faces/00027998.png")
img = rgb2gray(img)
img = img_as_ubyte(img)
img=img[:,:,np.newaxis]
#img = imread("data/processed_images/13_train_set_aflw/train/faces/001111.jpg")
for mul in xrange(3, 20):
im = resize(img, (mul*10, mul*10))
im = img_as_ubyte(im)
arr = skimage.util.view_as_windows(im, (13, 13, NUM_CHANNELS), step=STEP_SIZE)
f = file(LOAD_STATE_FROM_FILE, 'rb')
obj = pickle.load(f)
f.close()
arr = np.rollaxis(arr, 5, 3)
borrow = True
shared_x = theano.shared(np.asarray(arr, dtype=theano.config.floatX), # @UndefinedVariable
borrow=borrow)
iT = T.lscalar()
jT = T.lscalar()
x = T.tensor3("x")
layer0_input = x.reshape((1, NUM_CHANNELS, 13, 13))
net = twelve_net(layer0_input, None, relu, obj)
prediction = net.log_regression_layer.y_pred
py_x = net.log_regression_layer.p_y_given_x
test_model = theano.function(
[iT, jT],
[prediction, py_x, layer0_input],
givens={
x: shared_x[iT, jT, 0, :, :, :]
}
)
rows = arr.shape[0]
cols = arr.shape[1]
count = 0
faces = []
fig,axarr = plt.subplots(1,2)
confidence_map = np.zeros((rows,cols))
for i in xrange(rows):
for j in xrange(cols):
[y,p_y_given_x,f] = test_model(i,j)
f=f.reshape(NUM_CHANNELS,13,13)
f = np.rollaxis(f,0,3)
f = f[:,:,0]
confidence_map[i,j] = p_y_given_x[0,1]
# plt.imshow(f,cmap = "Greys_r")
# plt.show()
if y == 1:
count += 1
faces.append([i*STEP_SIZE,j*STEP_SIZE])
print i,j
print ("Check")
# hack for gray
im = im[:,:,0]
confidence_map=np.pad(confidence_map, 6,'constant')
axarr[0].get_xaxis().set_visible(False)
axarr[0].get_yaxis().set_visible(False)
axarr[1].get_xaxis().set_visible(False)
axarr[1].get_yaxis().set_visible(False)
axarr[1].imshow(confidence_map,cmap = "Greys_r")
axarr[0].imshow(im,cmap = "Greys_r")
img_desc = plt.gca()
#plt.imshow(im,cmap = "Greys_r")
#img_desc = plt.gca()
for point in faces:
topleftx = point[1]
toplefty = point[0]
rect = patches.Rectangle(
(topleftx, toplefty),
13,
13,
fill=False,
color='c'
)
axarr[0].add_patch(rect)
print count
fig_name = "lena_" + str(mul)+".png"
plt.savefig(fig_name, bbox_inches='tight')
def test_output_shape_arg_type(_type):
img = np.random.rand(3, 3)
output_shape = _type([5, 5])
assert resize(img, output_shape).shape == tuple(output_shape)
def test_downsize_anti_aliasing():
x = np.zeros((10, 10), dtype=np.double)
x[2, 2] = 1
scaled = resize(x, (5, 5), order=1, anti_aliasing=True, mode='constant')
assert scaled.shape == (5, 5)
assert np.all(scaled[:3, :3] > 0)
assert scaled[3:, :].sum() == 0
assert scaled[:, 3:].sum() == 0
sigma = 0.125
out_size = (5, 5)
resize(x,
out_size,
order=1,
mode='constant',
anti_aliasing=True,
anti_aliasing_sigma=sigma)
resize(x,
out_size,
order=1,
mode='edge',
anti_aliasing=True,
anti_aliasing_sigma=sigma)
resize(x,
out_size,
order=1,
mode='symmetric',
anti_aliasing=True,
anti_aliasing_sigma=sigma)
resize(x,
out_size,
order=1,
mode='reflect',
anti_aliasing=True,
anti_aliasing_sigma=sigma)
resize(x,
out_size,
order=1,
mode='wrap',
anti_aliasing=True,
anti_aliasing_sigma=sigma)
with pytest.raises(ValueError): # Unknown mode, or cannot translate mode
resize(x,
out_size,
order=1,
mode='non-existent',
anti_aliasing=True,
anti_aliasing_sigma=sigma)
