def load_segmentation(source_image_filename, mask_image_filename):
'''
Load Segmentation data.
Parameters
----------
source_image_filename : string filename of the source image
mask_image_filename : string filename of the corresponding mask image
in binary format
Notes
-----
In Segmentation model we have to feed the model with a simple image and
the labels will be given by the mask (binary) of the same image in which
the segmentation parts are highlight
No checks are performed on the compatibility between source image and
corresponding mask file.
The only checks are given on the image size (channels are excluded)
'''
src_image = Image(source_image_filename)
mask_image = Image(mask_image_filename)
if src_image.shape[:2] != mask_image.shape[:2]:
raise ValueError(
'Incorrect shapes found. The source image and the corresponding mask have different sizes'
)
return (src_image, mask_image)
def read (self):
'''
Get a frame as Image object
Returns
-------
im : Image obj
The loaded image
'''
im = Image()
return im.from_frame(self._queue.get())
def load_super_resolution(hr_image_filename, patch_size=(48, 48), scale=4):
'''
Load Super resolution data.
Parameters
----------
hr_image_filename : string
Filename of the high resolution image
patch_size : tuple (default=(48, 48))
Dimension to cut
scale : int (default=4)
Downsampling scale factor
Returns
-------
data : Image obj
Loaded Image object
label : Image obj
Generated Image label
Notes
-----
.. note::
In SR models the labels are given by the HR image while the input data are obtained
from the same image after a downsampling/resizing.
The upsample scale factor learned by the SR model will be the same used inside this
function.
'''
hr_image = Image(hr_image_filename)
w, h, _ = hr_image.shape
patch_x, patch_y = patch_size
dx = np.random.uniform(low=0, high=w - patch_x - 1)
dy = np.random.uniform(low=0, high=h - patch_y - 1)
hr_image = hr_image.crop(dsize=(dx, dy), size=patch_size)
random_flip = np.random.uniform(low=0, high=1.)
if random_flip >= .66:
hr_image = hr_image.transpose()
hr_image = hr_image.flip()
elif random_flip >= .33:
hr_image = hr_image.flip()
else:
pass
label = hr_image
data = hr_image.resize(scale_factor=(scale, scale))
return (data, label)
def main():
done = True
args = parse_args()
data_cfg = data_config(args.data_cfg)
args.netcfg = data_cfg.get('cfg', default=args.netcfg)
args.weights = data_cfg.get('weights', default=args.weights)
args.namesfile = data_cfg.get('names', default=args.namesfile)
if not args.netcfg or not args.weights:
raise ValueError('Network config AND network weights must be given')
names = get_labels(args.namesfile, args.classes)
net = Network(batch=32)
net.load(args.netcfg, args.weights)
net_w, net_h, _ = net.input_shape
if not args.input:
args.input = input('Enter Image Path: ')
done = False if not args.input else True
# set the output filename
args.outfile = args.outfile if args.outfile else splitext(
args.input)[0] + '_detected'
while done:
# load image from file
input_image = Image(filename=args.input)
# pad-resize image if it is necessary
input_image = input_image.letterbox(net_dim=(
net_w,
net_h)) if input_image.shape[:2] != net.shape else input_image
_ = net.predict(X=input_image)
# insert boxes evaluation and draw-detection and show image
input_image.show(window_name=args.outfile,
ms=0,
fullscreen=args.fullscreen)
if args.save:
input_image.save(filename=args.outfile)
args.input = input('Enter Image Path: ')
done = False if not args.input else True
def read(self):
'''
Get a frame as Image object
'''
im = Image()
return im.from_frame(self._queue.get())