This is a project for detection of the 19 cephalometric landmarks on X-ray images using semantic segmentation using Unet
First Trial The use of 1 channel binary mask where 1's represent a white square with the landmark position as the centre, and 0's represent the background Dice loss function was used
Second Trial (ongoing) The mask consists of 20 channels (Heatmaps) representing the 19 landmarks and the background and was filled with the below function
