mtuann / image-segmentation

image-segmentation

Related Paper for Segmentation

• Loss functions
• SOTA-Medical-Image-Segmentation-Update 2021-01
• Active Deep Learning for Medical Imaging Segmentation
• Awesome GAN for Medical Imaging
• VNet
• A 3D multi-modal medical image segmentation library in PyTorch
• UNet: semantic segmentation with PyTorch
• UNet-related model

Convolutional Layers:

Shape format:

• image: (h, w, c), filter: (f_h, f_w, c), padding: P = (F - 1)/ 2, F: width of the padding, stride: S
  • output_width = ((W-F) + 2 * P / S ) + 1
• Number of parameters in layers:
  • Convolution layers: feature dimension: l -> k: ( m * n * l + 1 ) * k; m, n is the shape of the Convolutional filter

Set up enviroments

1. Fill out the library in enviroment.yml file
2. Create condata enviroments with prefix conda env create --prefix ./env --file environment.yml # create the environment
3. Activate enviroments conda activate ./env # activate the environment
4. Check package list conda list
5. Check version of the package conda list | grep numpy
6. Update package conda env update --prefix ./env --file environment.yml --prune # update the environment
7. For more information about conda
8. Update torch with cuda

conda install pytorch torchvision torchaudio cudatoolkit=10.2 -c pytorch
nvcc --version
conda activate /data.local/linhlpv/envs/aicardio

Other check CUDA

1. cuda version: nvidia-smi
2. cudnn version: /usr/include/cudnn.h | grep CUDNN_MAJOR -A 2
3. cudatoolkit version: nvcc --version