def unnormalize_data(data, data_range, debug=True):
'''
this function unnormalizes 1-d label to normal scale based on range of data
'''
if debug:
print(
'debug mode is on during data unnormalizing. Please turn off after debuging'
)
assert isnparray(data), 'only numpy array is supported'
if istuple(data_range) or islist(data_range):
assert len(data_range) == 2, 'data range is not correct'
elif isnparray(data_range):
assert data_range.size == 2, 'data range is not correct'
max_value = data_range[1]
min_value = data_range[0]
unnormalized = data * (max_value - min_value) + min_value
if debug:
normalized = normalize_data(data=unnormalized,
data_range=data_range,
debug=False)
assert_almost_equal(data,
normalized,
decimal=6,
err_msg='data is not correct: %f vs %f' %
(data, normalized))
return unnormalized
def normalize_data(data, data_range=None, debug=True):
'''
this function normalizes 1-d label to 0-1
'''
if debug:
assert isnparray(data), 'only numpy array is supported'
print(
'debug mode is on during data normalizing. Please turn off after debuging'
)
if data_range is None:
max_value = np.max(data)
min_value = np.min(data)
else:
if debug:
if istuple(data_range) or islist(data_range):
assert len(data_range) == 2, 'data range is not correct'
elif isnparray(data_range):
assert data_range.size == 2, 'data range is not correct'
max_value = data_range[1]
min_value = data_range[0]
normalized_data = float(data - min_value)
normalized_data = normalized_data / (max_value - min_value)
if debug:
unnormalized = unnormalize_data(data=normalized_data,
data_range=(min_value, max_value),
debug=False)
assert_almost_equal(data,
unnormalized,
decimal=6,
err_msg='data is not correct: %f vs %f' %
(data, unnormalized))
return normalized_data
def apply_rotation_tight(bbox_in, angle_in_degree, im_shape, debug=True):
'''
return 4 points clockwise
'''
if debug:
assert isnparray(bbox_in) and bbox_in.size == 4, 'box is not correct'
bbox_in = np.reshape(bbox_in, (4, ))
bbox_tight = bbox_rotation_inv(bbox_in, angle_in_degree, im_shape, debug=debug) # get top left and bottom right coordinate of the rotated bbox in the image coordinate
# print('bbox after inverse the rotation')
# print(bbox_tight)
pts_total = np.zeros((4, 2), dtype=np.int)
pts_tl = np.array([bbox_tight[0], bbox_tight[1]])
pts_br = np.array([bbox_tight[2], bbox_tight[3]])
line1 = get_line(imagecoor2cartesian(pts_tl, debug=debug), angle_in_degree + 90.00, debug=debug)
line2 = get_line(imagecoor2cartesian(pts_br, debug=debug), angle_in_degree, debug=debug)
pts_bl = cartesian2imagecoor(get_intersection(line1, line2, debug=debug), debug=debug)
pts_tr = cartesian2imagecoor(get_intersection(get_line(imagecoor2cartesian(pts_tl, debug=debug), angle_in_degree, debug=debug), get_line(imagecoor2cartesian(pts_br, debug=debug), angle_in_degree + 90.00, debug=debug), debug=debug), debug=debug)
# print np.reshape(pts_tl, (1, 2)).shape
# print pts_total[0, :].shape
pts_total[0, :] = np.reshape(pts_tl, (1, 2))
pts_total[1, :] = np.reshape(pts_tr, (1, 2))
pts_total[2, :] = np.reshape(pts_br, (1, 2))
pts_total[3, :] = np.reshape(pts_bl, (1, 2))
return pts_total
def bbox_rotatedtight2rotatedloose(bbox_in, angle_in_degree, debug=True):
'''
transfer the rotated bbox with tight version to loose version, both contains only two points (top left and bottom right)
only a single box is feeded into
'''
if debug:
assert isnparray(bbox_in) and bbox_in.size == 4, 'box is not correct'
pts_tl = np.array([bbox_in[0], bbox_in[1]])
pts_br = np.array([bbox_in[2], bbox_in[3]])
line1 = get_line(imagecoor2cartesian(pts_tl), angle_in_degree + 90.00)
line2 = get_line(imagecoor2cartesian(pts_br), angle_in_degree)
pts_bl = cartesian2imagecoor(get_intersection(line1, line2))
pts_tr = cartesian2imagecoor(get_intersection(get_line(imagecoor2cartesian(pts_tl), angle_in_degree), get_line(imagecoor2cartesian(pts_br), angle_in_degree + 90.00)))
# assert_almost_equal(np.dot(pts_bl - pts_br, pts_bl - pts_tl), 0, err_msg='The intersection points are wrong')
# assert_almost_equal(np.dot(pts_tr - pts_br, pts_tr - pts_tl), 0, err_msg='The intersection points are wrong')
pts_tl_final = np.zeros((2), dtype=np.float32)
pts_br_final = np.zeros((2), dtype=np.float32)
pts_tl_final[0] = min({pts_tl[0], pts_br[0], pts_bl[0], pts_tr[0]})
pts_tl_final[1] = min({pts_tl[1], pts_br[1], pts_bl[1], pts_tr[1]})
pts_br_final[0] = max({pts_tl[0], pts_br[0], pts_bl[0], pts_tr[0]})
pts_br_final[1] = max({pts_tl[1], pts_br[1], pts_bl[1], pts_tr[1]})
# print(pts_tl_final)
# print(pts_br_final)
test = np.hstack((pts_tl_final, pts_br_final))
return test
def identity(data, data_range=None, debug=True):
if debug:
print(
'debug mode is on during identity function. Please turn off after debuging'
)
assert isnparray(data), 'data is not correct'
return data
def print_np_shape(nparray, debug=True):
'''
print a string to represent the shape of a numpy array
'''
if debug:
assert isnparray(
nparray), 'input is not a numpy array and does not have any shape'
return '(%s)' % (functools.reduce(lambda x, y: str(x) + ', ' + str(y),
nparray.shape))
def bboxcheck(bbox, debug=True):
'''
check the input to be a bounding box
parameter:
bbox: N x 4 numpy array, N >= 0
return:
True or False
'''
return isnparray(bbox) and bbox.shape[1] == 4 and bbox.shape[0] >= 0
def bbox_rotation_inv(bbox_in, angle_in_degree, image_shape, debug=True):
'''
bbox_in is two coordinate
angle is clockwise
'''
if debug:
assert isnparray(bbox_in) and bbox_in.size == 4, 'box is not correct'
im_width = image_shape[1]
im_height = image_shape[0]
coor_in_tl = np.array([(bbox_in[0] - im_width/2)/im_width*2, (bbox_in[1] - im_height/2)/im_height*2, 1]) # normalization
coor_in_br = np.array([(bbox_in[2] - im_width/2)/im_width*2, (bbox_in[3] - im_height/2)/im_height*2, 1]) # normalization
# print(coor_in_tl)
# print(coor_in_br)
affine = np.array([[math.cos(rad(angle_in_degree)), math.sin(rad(angle_in_degree)), 0], [-math.sin(rad(angle_in_degree)), math.cos(rad(angle_in_degree)), 0]])
coor_out_tl = np.dot(coor_in_tl, affine.transpose())
coor_out_br = np.dot(coor_in_br, affine.transpose())
# print(coor_out_tl)
# print(coor_out_br)
bbox_recover = [coor_out_tl[0] * im_width/2 + im_width/2, coor_out_tl[1] * im_height/2 + im_height/2, coor_out_br[0] * im_width/2 + im_width/2, coor_out_br[1] * im_height/2 + im_height/2]
bbox_recover = np.array(bbox_recover, dtype = float)
return bbox_recover
def unpreprocess_image_caffe(image_datablob,
pixel_mean=None,
swap_channel=True,
debug=True):
'''
this function unpreprocesses image for caffe only,
including transfer from bgr to rgb
from NxCxHxW to a list of HxWxC
'''
if debug:
print(
'debug mode is on during unpreprocessing. Please turn off after debuging'
)
assert isnparray(image_datablob
) and image_datablob.ndim == 4, 'input is not correct'
assert image_datablob.shape[1] == 1 or image_datablob.shape[
1] == 3, 'this is not an blob of image, channel is not 1 or 3'
if pixel_mean is not None:
assert pixel_mean.shape == (1, 1, 3) or pixel_mean.shape == (
1, ), 'pixel mean is not correct'
pixel_mean_reshape = np.reshape(pixel_mean, (1, 3, 1, 1))
image_datablob += pixel_mean_reshape
image_datablob = np.transpose(
image_datablob,
(0, 2, 3, 1)) # permute to [batch, height, weight, channel]
if image_datablob.shape[-1] == 3 and swap_channel: # channel dimension
image_datablob = image_datablob[:, :, :,
[2, 1,
0]] # from bgr to rgb for color image
image_data_list = list()
for i in xrange(image_datablob.shape[0]):
image_data_list.append(image_datablob[i, :, :, :])
return image_data_list
