Generic Foreground Segmentation in Images
Python Matlab
Switch branches/tags
Nothing to show
Clone or download
Latest commit 662b354 Jan 23, 2017
Failed to load latest commit information.
images Pixel objectness Jan 20, 2017 Updated Readme Jan 23, 2017
blend_mask.m Pixel objectness Jan 20, 2017 Updated demo Jan 22, 2017
pixel_objectness.caffemodel Pixel objectness Jan 20, 2017
show_results.m Pixel objectness Jan 20, 2017
test_template.prototxt Pixel objectness Jan 20, 2017

Pixel Objectness

The following repository contains pretrained model for pixel objectness.

Please visit our project page for the paper and visual results.

If you use this in your research, please cite the following paper:

  Author = {Jain, Suyog and Xiong, Bo and Grauman, Kristen},
  Journal = {arXiv preprint arXiv:1701.05349},
  Title = {Pixel Objectness},
  Year = {2017}

These models are freely available for research and academic purposes. However it's patent pending, so please contact us for any commercial use.

Using the pretrained models:

This model is trained using Deeplab-v1 caffe library. Please cite [1] and [2] if you use the code.

  • Setup: Download and install Deeplab-v1 from here

  • Refer to for step-by-step instruction on how to run the code.

  • Store the images that you want to process in the images folder.

  • Update the caffe binary path and image extension variable in

  • Running will produce three files 1) image_list.txt : contains list of of input images, 2) output_list.txt: contains names to be used to store the output of pixel objectness 3) test.protoxt: prototxt file required for loading the pretrained model.

  • Please resize your images so that the maximum side is < 513, otherwise update the crop_size value in test_template.prototxt. Bigger crop sizes require larger gpu memory.

Visualizing the results:

After execution will store pixel objectness results as matlab files.

Please refer to show_results.m to see how to visualize and extract foreground masks.

Please cite these too if you use the code:

[1] Caffe:

Author = {Jia, Yangqing and Shelhamer, Evan and Donahue, Jeff and Karayev, Sergey and Long, Jonathan and Girshick, Ross and Guadarrama, Sergio and Darrell, Trevor},
Journal = {arXiv preprint arXiv:1408.5093},
Title = {Caffe: Convolutional Architecture for Fast Feature Embedding},
Year = {2014}

[2] Deeplab-v1:

title={Semantic Image Segmentation with Deep Convolutional Nets and Fully Connected CRFs},
author={Liang-Chieh Chen and George Papandreou and Iasonas Kokkinos and Kevin Murphy and Alan L Yuille},