Skip to content

Latest commit

 

History

History
executable file
·
22 lines (12 loc) · 985 Bytes

README.md

File metadata and controls

executable file
·
22 lines (12 loc) · 985 Bytes

DL SkolTech project.

Cascade Neural Networks Algorithms for Segmentation in Medical Imaging.

The final report for the project can be found here.

Github reps that were used:

2d Unet: https://github.com/cosmic-cortex/pytorch-UNet

3d Unet: https://github.com/wolny/pytorch-3dunet

psp net (pspnet was used like carcase and was modified for our task): https://github.com/Lextal/pspnet-pytorch/blob/master/pspnet.py

encoders: https://github.com/qubvel/segmentation_models.pytorch

encoders: https://github.com/Cadene/pretrained-models.pytorch#resnext

Spatial and channel-wise attention: https://github.com/LeeJunHyun/Image_Segmentation, https://github.com/ozan-oktay/Attention-Gated-Networks/blob/master/models/networks/unet_grid_attention_3D.py

FPN: https://github.com/qubvel/segmentation_models.pytorch

all of that githubs were modified for our tasks and our pipeline.