SegNet including indices pooling for Semantic Segmentation with tensorflow and keras
Branch: master
Clone or download
Latest commit 591c8bd Oct 30, 2018
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
.gitignore Deleted *.ipynb Oct 29, 2017
README.md Fix typo Sep 25, 2018
generator.py Flake8 Oct 30, 2018
layers.py Flake8 Oct 30, 2018
model.py Flake8 Oct 30, 2018
train.py Flake8 Oct 30, 2018

README.md

SegNet

SegNet is a model of semantic segmentation based on Fully Comvolutional Network.

This repository contains the implementation of learning and testing in keras and tensorflow. Also included is a custom layer implementation of index pooling, a new property of segnet.

architecture

  • encoder decoder architecture

  • fully convolutional network

  • indices pooling

    indicespooling

description

This repository is SegNet architecture for Semantic Segmentation. The repository of other people's segmentation, pooling with indices not implemented.But In this repository we implemented pooling layer and unpooling layer with indices at MyLayers.py.

Segnet architecture is early Semantic Segmentation model,so acccuracy is low but fast. In the future, we plan to implement models with high accuracy.(UNet,PSPNet,Pix2Pix ect..)

Usage

train

python SegNet.py [--options your dataset]

DEMO

  • dataset
    • LIP(Look Into Person)

The problem of binary mask works well with SegNet.

demo2