Create README.md

README.md

@@ -0,0 +1 @@
+I went with the approach of training a new model on a dataset where the segmentation maps, instead of being a probability map between 0 and 1, to give each KP a value of 1, 2, 3 etc