def build_pairs(x, y, epochs):
stream_1 = ds.epochs(zip(x, y), epochs=len(y))
stream_1 = ds.epochs(stream_1, epochs=epochs, random=False)
stream_2 = repeat_elem(zip(x, y), times=len(y))
stream_2 = ds.epochs(stream_2, epochs=epochs, random=False)
stream = skip_equals(stream_1, stream_2)
return stream
def prepare(dataset, input_shape):
stream = ds.epochs(dataset.image_ids, epochs=1)
# stream = ds.stream(check, stream)
stream = ds.stream(
lambda x: (dataset._img_filenames[x], dataset.load_image(x)), stream)
stream = ds.stream(ds.apply_to_y(ds.resize(input_shape)), stream)
return stream
def prepare(dataset, input_shape, output_shape):
resizer = ds.resize(u.mul_shape(output_shape, 32))
stream = ds.epochs(dataset.image_ids, epochs=1)
stream = ds.stream(
lambda x: (dataset._img_filenames[x], resizer(dataset.load_image(x))),
stream)
return stream
def get_dataset(filenames, epochs, colorMap, batch_size, output_shape):
shapes = load_shapes(ds.epochs(filenames, epochs), colorMap)
shapes = ds.stream(ds.apply_to_x(lambda x: cv2.resize(x, input_shape[:2])),
shapes)
shapes = ds.stream(
ds.apply_to_y(lambda x: to_categorical(x, output_shape)), shapes)
shapes = ds.stream_batch(
shapes, lambda x, y: [np.array(x), np.array(y)], batch_size)
return shapes
def prepare(dataset, epochs, batch_size, base_path):
get_masks = load_masks(dataset, base_path)
stream = ds.epochs(dataset.image_ids, epochs)
stream = ds.stream(lambda x: (dataset.load_image(x), get_masks(x)), stream)
stream = ds.bufferize(stream, size=batch_size)
batch = ds.stream_batch(stream, size=batch_size, fun=ds.pack_elements)
return batch
def prepare(dataset, epochs, batch_size, input_shape, base_path):
lm = load_masks(dataset, base_path)
stream = ds.epochs(dataset.image_ids, epochs)
stream = ds.stream(lambda x: (dataset.load_image(x), lm(x)), stream)
# stream = ds.stream(ds.apply_to_x(ds.resize(input_shape)), stream)
batch = ds.stream_batch(stream,
size=batch_size,
fun=lambda x, y: (np.array(x), fix_y(y)))
return batch
def prepare(dataset, epochs, batch_size, input_shape, output_shape):
i2c = imageid2categoricalpixel(dataset, input_shape)
stream = ds.epochs(dataset.image_ids, epochs)
stream = ds.stream(lambda x: (dataset._img_filenames[x], i2c(x)), stream)
def adapt_y(x, y):
y = np.array(y).transpose((2, 0, 1))
y = y.reshape(y.shape + (1, )) # .astype('int32')
return np.array(x), [v for v in y]
batch = ds.stream_batch(stream, size=batch_size, fun=adapt_y)
return batch
def get_dataset(filenames, relations, epochs, colorMap, batch_size,
input_shape, output_shape):
shapes = load_shapes(ds.epochs(filenames, epochs), relations, colorMap)
shapes = ds.stream(ds.apply_to_x(lambda x: cv2.resize(x, input_shape[:2])),
shapes)
shapes = ds.stream(
ds.apply_to_y(lambda x: np.array(
[to_categorical(xv, output_shape) for xv in x]).sum(axis=0)),
shapes)
shapes = ds.stream_batch(
shapes,
lambda x, y: [np.array(x), generate_y(np.array(y))], batch_size)
return shapes
def prepare(dataset, epochs, batch_size, input_shape, output_shape):
stream = ds.epochs(dataset.image_ids, epochs)
stream = ds.stream(
lambda x: (dataset.load_image(x), dataset.load_output(x)),
stream)
stream = ds.stream(ds.apply_to_x(ds.resize(input_shape)), stream)
stream = ds.stream(ds.apply_to_y(resize_all(output_shape)), stream)
stream = ds.bufferize(stream, size=20)
batch = ds.stream_batch(stream, size=batch_size, fun=ds.pack_elements)
batch = ds.stream(ds.apply_to_y(ds.apply_to_xn(
lambda x: ds.image2mask(x).reshape(x.shape + (1,)))), batch)
return batch
def get_dataset(dataset, epochs, input_shape, output_shape, colored=True):
shapes = ds.epochs(dataset, epochs)
shapes = ds.stream(
ds.apply_to_x(
ds.apply_to_xn(lambda x: ds.colorize(
cv2.imread(x[0]), x[1] if colored else [255, 0, 0]))), shapes)
# shapes = ds.stream(ds.apply_to_x(check), shapes)
shapes = ds.stream(ds.apply_to_x(sum), shapes)
shapes = ds.stream(ds.apply_to_x(lambda x: cv2.resize(x, input_shape[:2])),
shapes)
shapes = ds.stream(
ds.apply_to_y(lambda x: to_categorical(x, output_shape)), shapes)
shapes = ds.stream_batch(
shapes, lambda x, y: [np.array(x), np.array(y)], batch_size)
return shapes
def prepare(dataset, epochs, batch_size, input_shape):
stream = ds.epochs(dataset.image_ids, epochs)
stream = ds.stream(
lambda x: (dataset.load_image(x), dataset.load_output(x)), stream)
stream = ds.stream(ds.apply_to_xn(ds.resize(input_shape[:2])), stream)
# stream = ds.stream(ds.apply_to_x(check), stream)
stream = ds.bufferize(stream, size=20)
batch = ds.stream_batch(stream, size=batch_size)
batch = ds.stream(
ds.apply_to_y(lambda x: ds.mask2image(x).reshape(x.shape + (1, ))),
batch)
return batch
def prepare(dataset, epochs, batch_size, input_shape, output_shape, load_image,
resizer):
stream = ds.epochs(dataset.image_ids, epochs)
stream = ds.stream(
lambda x: (load_image(x), resizer(dataset.load_output(x))), stream)
def transpose(x, y):
y = np.array(y)
return np.array(x), y.reshape((1, ) + y.shape)
batch = ds.stream_batch(stream, size=batch_size, fun=transpose)
batch = ds.stream(
ds.apply_to_y(
ds.apply_to_xn(lambda x: ds.image2mask(x).reshape(x.shape +
(1, )))), batch)
return batch
def prepare(dataset, input_shape, output_shape):
stream = ds.epochs(dataset.image_ids, epochs=1)
stream = ds.stream(
lambda x: (dataset._img_filenames[x], (x, dataset.load_output(x))),
stream)
# stream = ds.stream(ds.apply_to_x(ds.resize(input_shape)), stream)
stream = ds.stream(ds.apply_to_y(resize_all(dataset, output_shape)),
stream)
stream = ds.bufferize(stream, size=10)
batch = ds.stream_batch(stream, size=10, fun=ds.pack_elements)
# batch = ds.stream(ds.apply_to_y(check), batch)
batch = ds.stream(
ds.apply_to_y(ds.apply_to_xn(lambda x: ds.image2mask(x))), batch)
return batch
def get_dataset(filenames, relations, epochs, colorMap, batch_size,
input_shape, output_shape):
shapes = load_shapes(ds.epochs(filenames, epochs), relations, colorMap)
shapes = ds.stream(ds.apply_to_x(lambda x: cv2.resize(x, input_shape[:2])),
shapes)
# def fu(x):
# ba = list(np.array(
# [to_categorical(xv, output_shape) for xv in x]
# ).sum(axis=0))
# import ipdb; ipdb.set_trace()
# return ba
shapes = ds.stream(
ds.apply_to_y(lambda x: np.array(
[to_categorical(xv, output_shape) for xv in x]).sum(axis=0)),
shapes)
shapes = ds.stream(ds.apply_to_y(generate_neg), shapes)
shapes = ds.stream_batch(
shapes, lambda x, y: [np.array(x), np.array(y)], batch_size)
return shapes
