Encoder-decoder networks

Last updated Jul 28, 2022 Edit Source

AI/Deep learning/DL architectures composed of two paths, an encoding and a decoding one. AI/Deep learning/Autoencoders are similar but unsupervised (reconstructions loss). U-NETs are a type of encoder-decoder AI/Deep learning/CNNs model with skipped connections trained in a AI/Supervised Learning/Supervised learning context for image segmentation and related tasks. Very common models for semantic segmentation tasks

• https://www.slideshare.net/PetteriTeikariPhD/multiphoton-vasculature-segmentation-5-unet

• #CODE Keras-unet-collection

• #PAPER U-Net: Convolutional Networks for Biomedical Image Segmentation (Ronneberger 2015)
  • https://towardsdatascience.com/review-u-net-biomedical-image-segmentation-d02bf06ca760
  • https://github.com/karolzak/keras-unet
  • https://tuatini.me/practical-image-segmentation-with-unet/
  • https://github.com/divamgupta/image-segmentation-keras/blob/master/keras_segmentation/models/unet.py
  • diagram example
• #PAPER 3D U-Net: Learning Dense Volumetric Segmentation from Sparse Annotation (Cicek 2016)
  • https://towardsdatascience.com/review-3d-u-net-volumetric-segmentation-medical-image-segmentation-8b592560fac1
• #PAPER V-Net: Fully Convolutional Neural Networks for Volumetric Medical Image Segmentation (Milletari 2016)
  • https://towardsdatascience.com/review-v-net-volumetric-convolution-biomedical-image-segmentation-aa15dbaea974
• #PAPER Volumetric ConvNets with Mixed Residual Connections for Automated Prostate Segmentation from 3D MR Images, 3D UNET+Resnet (Yu 2017)
  • https://towardsdatascience.com/review-3d-u-net-resnet-volumetric-convolutions-long-short-residual-connections-biomedical-3a7da3f98dae
• #PAPER Automatic 3D Cardiovascular MR Segmentation with Densely-Connected Volumetric ConvNets, DenseVoxNet (Yu 2017)
  • https://medium.com/@sh.tsang/review-densevoxnet-volumetric-brain-segmentation-biomedical-image-segmentation-9136bb6128dd
• #PAPER Road Extraction by Deep Residual U-Net, ResUNET (Zhang 2017)
  • https://github.com/nikhilroxtomar/Deep-Residual-Unet
• #PAPER Recurrent Residual Convolutional Neural Network based on U-Net (R2U-Net) for Medical Image Segmentation (Alom 2018)
• #PAPER UNet++: A Nested U-Net Architecture for Medical Image Segmentation (Zhou 2018)
  • https://medium.com/@sh.tsang/review-unet-a-nested-u-net-architecture-biomedical-image-segmentation-57be56859b20
• #PAPER H-DenseUNet: Hybrid Densely Connected UNet for Liver and Tumor Segmentation from CT Volumes (Li 2018)
  • https://medium.com/@sh.tsang/review-h-denseunet-2d-3d-denseunet-for-intra-inter-slice-features-biomedical-image-f3e526e81fe7
  • #CODE https://github.com/xmengli999/H-DenseUNet/
• #PAPER Encoder-Decoder with Atrous Separable Convolution for Semantic Image Segmentation, DeepLabv3+ (Chen 2018)
  • #CODE https://github.com/ChoiDM/pytorch-deeplabv3plus-3D
• #PAPER ResUNet-a: a deep learning framework for semantic segmentation of remotely sensed data (Diakogiannis 2019)
• #PAPER ResUNet++: An Advanced Architecture for Medical Image Segmentation (Jha 2019)
• #PAPER M-Net: U-Net with Multi-stream Feature Fusion and Multi-scale Dilated Convolutions (Fu 2019)
  • https://www.researchgate.net/publication/336455527_M-Net_A_Novel_U-Net_with_Multi-stream_Feature_Fusion_and_Multi-scale_Dilated_Convolutions_for_Bile_Ducts_and_Hepatolith_Segmentation_September_2019
• #PAPER Channel-Unet: A Spatial Channel-Wise Convolutional Neural Network for Liver and Tumors Segmentation (Chen 2019)
• #PAPER Bi-Directional ConvLSTM U-Net with Densely Connected Convolutions (Azad 2019)
  • #CODE https://github.com/rezazad68/BCDU-Net/blob/master/Retina%20Blood%20Vessel%20Segmentation/models.py
• #PAPER LSTM-UNET - Microscopy Cell Segmentation via Convolutional LSTM Networks (Arbelle 2019) #CODE https://github.com/arbellea/LSTM-UNet
• #PAPER USE-Net: incorporating Squeeze-and-Excitation blocks into U-Net for prostate zonal segmentation of multi-institutional MRI datasets (Rundo 2019)
• #PAPER Evaluation of Multi-Slice Inputs to Convolutional Neural Networks for Medical Image Segmentation (Vu 2019)
• #PAPER Making a Case for 3D Convolutions for Object Segmentation in Videos (Mahadevan 2020)
  • proposed a simple and fast network architecture consisting entirely of 3D convolutions that is capable of effectively learning spatio-temporal features
  • used a 3D ResNet pretrained for video action classification as an encoder, and a novel decoder architecture inspired by existing 2D convolutional networks
• #PAPER nnU-Net: a self-configuring method for deep learning-based biomedical image segmentation (Isensee 2020)
  • https://www.sciencedaily.com/releases/2020/12/201207112253.htm
  • #CODE https://github.com/MIC-DKFZ/nnUNet
• #PAPER MCNN, Multi-resolution convolutional neural networks for inverse problems (Wang 2020)
  • #CODE https://github.com/fengwang/MCNN
  • #CODE https://github.com/fengwang/mcnn-demo/tree/master/demo