Development of ACE-NET deep-neural-network based on ArXiv paper

The purpose of this method focus on highlighting changes between satellites' captured images. A special neural network architecture translates the images between the two domains of different remote sensors.

Uses Flood_UiT_HCD_California_2017_Luppino as training database.

Test the model using python main.py -c checkpoints/epoch=249-step=21999.ckpt --patch_size 250 --verbose

-c refers to path of the file with model's pretrained parameters. Train the network on your own or use one of checkpoints provided inside this repository.

part of Degree thesis