Automatic Deep Extreme Cut

Prerequisites

• Python 3
• Pytorch 0.3.6
• torchvision
• visdom (optional for visualization)

Installation

The code was tested with Anaconda and Python 3.6. After installing the Anaconda environment:

0. Clone the repo:

   git clone https://github.com/kh22l22/Auto-DEXTR-Pytorch
   cd Auto-DEXTR-Pytorch

1. Install dependencies:

   conda install pytorch torchvision -c pytorch
   conda install matplotlib opencv pillow scikit-learn scikit-image

2. Run demo code

   python demo.py

3. default choose: download and copy the pretrained model to weights directory.

Related Projects:

NLFD: Luo Z, Mishra A, Achkar A, Eichel J, Li S-Z, Jodoin P-M, “Non-Local Deep Features for Salient Object Detection”, IEEE Conference on Computer Vision and Pattern Recognition (CVPR 2017)

Deep Extreme Cut: K.K. Maninis*, S. Caelles*, J. Pont-Tuset, and L. Van Gool Deep Extreme Cut: From Extreme Points to Object Segmentation, Computer Vision and Pattern Recognition (CVPR), 2018.

Acknowledgments

Code borrows from NLFD and Deep Extreme Cut.