Upload QES.py

other scripts for the dataset's construction/QES.py

@@ -0,0 +1,19 @@
+# The input of QES should be a soft mask prediction instead of a binarized one.
+
+import cv2
+import argparse
+import numpy as np
+
+parser = argparse.ArgumentParser()
+parser.add_argument("--mask", required=True, type=str, help="The path of the input mask.")
+args = parser.parse_args()
+
+t1 = (1 / 16)
+t2 = (15 / 16)
+
+mask = (cv2.imread(args.mask, 0).astype(np.float32)/255.0)
+QES = ((mask>t2).sum()/((mask>t1).sum()+1e-8))
+
+print('The Quality Evaluation Score is:', QES)
+
+