- U-Net: Convolutional Networks for Biomedical Image Segmentation
- FusionNet: A deep fully residual convolutional neural network for image segmentation in connectomics
- Pytorch 0.1.12
- Python 3.5.2
- wget
git clone https://github.com/GunhoChoi/Kind_PyTorch_Tutorial.git
cd Kind_PyTorch_Tutorial/13_Semantic_Segmentation/
wget https://people.eecs.berkeley.edu/~tinghuiz/projects/pix2pix/datasets/maps.tar.gz
tar -xzvf maps.tar.gz
mkdir model result
parser.add_argument("--network",type=str,default="fusionnet",help="choose between fusionnet & unet")
parser.add_argument("--batch_size",type=int,default=1,help="batch size")
parser.add_argument("--num_gpu",type=int,default=1,help="number of gpus")
python3 main.py --network unet --batch_size 1 --num_gpu 1
Original Image / Label Image / Generated Image
