wuzuowuyou / pytorch_semantic_segmentation

Implement some models of RGB/RGBD semantic segmentation in PyTorch, easy to run. Such as FCN, RefineNet, PSPNet, RDFNet, 3DGNN, PointNet, DeepLab V3, DeepLab V3 plus, DenseASPP, FastFCN

PyTorch_Semantic_Segmentation

Implement some models of semantic segmentation in PyTorch, easy to run.

Introduction

1. This repo includes some networks for Semantic Segmentation implemented in pytorch 1.0.0 and python3. See each directory for more information.
2. I only provide architecture of network here. Dataset and train/test files aren't available here, for I think it can be added according to individual needs.
3. The code file containing the network structure can be run directly with the set simulation data.
4. ResNet101 used in this repo is the one which PSPNet used. The difference between this resnet and the original resnet is that the first 7*7 conv layer in the old version is replaced by three small-kernel convs. Pretrained model can be downloaded from here.

Has finished

FCN8s

RefineNet (CVPR 2017)

PSPNet (CVPR 2017)

PointNet (CVPR 2017)

RDFNet (ICCV 2017)

3DGNN (ICCV 2017)

DeepLab V3

DeepLab V3+ (ECCV 2018)

DenseASPP (CVPR 2018)

FastFCN (Arxiv 2019)