Wing Landmark Detection model with image segmentation and blob detection. Used UNet model architecture to train the model.
U-Net was used to handle the segmentation tasks.The network aims to process the wing images and output a mask image that has the same size as the input image and consists of pixels that are in the range of 0-1. Annotated wing image coordinates are used to create landmark point masks. After that, those masks were used in training the model.
After segmentation, Blob detection method is used to extract coordinates of the landmarks.
