Tensorflow implementation of Dilated Residual Network from Fisher et al. 2017. Contains both 18 layer and 26 layer implementations for semantic segmentation. Cross-entropy, weighted cross-entropy and DICE loss can be used.
Users should enter values for batch_size, num_epochs, image_dims and num_classes in train_network.py
Data is imported using the tensorflow Dataset API. A separate file for both train and validation/test data should be generated. Each line in the file should contain the path to each image and it's label and weight
data_path/path_to_train_image data_path/path_to_label_image data_path/path_to_weight_image
