def process_and_evaluate(data):
filename_in = data["filename_in"]
filename_out = data["filename_out"]
evaluator = data["evaluator"]
processor = data["processor"]
print("Processing file: {}".format(filename_in))
# Load image
sample = Sample(filename_in, crop_bottom=0.10)
# Compute segmentation
final_img = processor.process(img=sample.get_image())
# Compute error
s, _ = evaluator.evaluate(final_img)
# dump segmentation
print("Dumping file: {} - MAE: {}".format(filename_out, math.fabs(s)))
cv2.imwrite(filename_out, final_img)
return s
default=False,
help="Put True if background is white")
parser.add_argument("-invert_grain",
type=bool,
default=False,
help="Put True if background is white")
args = parser.parse_args()
######
# Step 1 : Align
#####
# Look for the ref segment
segment = Sample(args.seg_ref_path)
segment = segment.get_image()
if args.invert_segment:
segment = np.invert(
segment
) # we need the background to be black (default color for numpy transformation)
segment[segment < 128] = 0
segment[segment >= 128] = 255
# Load gains
grain = Sample(args.grain_ref_path)
grain = grain.get_image()
if args.invert_grain:
grain = np.invert(
