Skip to content

This repository contains the entire pipline (including data preprocessing, training, testing, evaluation and visualization) for the Shearlet-based Semantic Edge Detection, .

License

arsenal9971/shearlet_semantic_edge

master
Switch branches/tags

Name already in use

A tag already exists with the provided branch name. Many Git commands accept both tag and branch names, so creating this branch may cause unexpected behavior. Are you sure you want to create this branch?
Code

Latest commit

 

Git stats

Files

Permalink
Failed to load latest commit information.
Type
Name
Latest commit message
Commit time
 
 
 
 
 
 
 
 
 
 

ShearSED: Shearlet Semantic Edge Detection

By H. Andrade-Loarca, G. Kutyniok, O. Öktem

Architecture DeNSE

Phantom results

Architecture shearCASENet

ShearCASENet results

Architecture shearDDS

ShearDDS results

License

ShearSED is released under the MIT License (refer to the LICENSE file for details).

Contents

  1. Introduction
  2. Citation
  3. Requirements
  4. Installation
  5. Usage
  6. References
  7. Contact

Introduction

This repository contains the entire pipline (including data preprocessing, training, testing, evaluation and visualization) for ShearSED.

ShearSED is a recently proposed set techniques which which involved model-based and data-driven approaches for high-performance semantic edge detection. This method uses the optimal edge representation in images provided by Shearlets and the highly specilized and accurate classification capabilities of deep convolutional neural networks. For more details, please refere to the arXiv technical report.

This work makes use of three different architectures, the first architecture uses a simple 4-layered CNN, also known as the DeNSE architecture, to extract the Wavefront set of an image. The method proposed earlier this year by the authors, DeNSE achieves state-of-the-art edge orientation extraction performance on the Semantic Boundaries Dataset and the Berkeley Segmentation Dataset, as well as other toy dataset with phatoms formed by ellipses which resembles human-head phantoms (e.g. Shepp-Logan phantom).

The other architectures, namely, ShearCASENet and ShearDDS, are able to perform semantic edge detection. This architectures use as the backbone the CASENet and DDS, but they take as input the shearlet coefficients of the image and it does not make use of the buffer layer of the original architectures.

The implementation for the shearCASENet and shearDDS is based on the pytorch implemenation of the original CASENet architecture provided by @lijiaman.

This method can be used for different applications in image processing and computer vision (e.g. edge/corner detection and tracking) as well as inverse problems regularization (e.g. Wavefront set reconstrucion in Computed Tomography).

Citation

If you find ShearSED useful in your research, please consider to cite the following papers:

@inproceedings{andrade2019sed, 
  title={Shearlets as Feature Extractor for Semantic Edge Detection: The Model-Based and Data-Driven Realm}, 
  author={Andrade-Loarca, Hector, Kutyiniok, Gitta, Öktem, Ozan},
  booktitle={arXiv preprint: arXiv:1911.12159}, 
  year={2019}
}

Requirements

Installation

Usage

References

Contact

Hector Andrade-Loarca

Questions can also be left as issues in the repository. We will be happy to answer them.

About

This repository contains the entire pipline (including data preprocessing, training, testing, evaluation and visualization) for the Shearlet-based Semantic Edge Detection, .

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages