GML22/GML-Net

[This project was completed in September 2020] The GML-Net is a convolutional neural network (CNN) that is based on U-Net architecture with an encoder derived from the ResNet family and BottleNeck blocks that provide reading and aggregation of feature maps from a cross-section of various scales. Effective network learning is ensured by loss function defined as a weighted sum of Binary Cross-Entropy Loss, Dice Loss and Lovász hinge Loss.

Jupyter NotebookMIT