GIST-feature-extractor

[UnOfficial] This repository is a reimplementation Matlab code implemented in this paper Modeling the shape of the scene: a holistic representation of the spatial envelope in Python.

Note : Images that can be used in this repository are only grayscale images. (W × H × 1)

Prerequisites

The folloing liblaries are required to be installed

• Python 3.6
• numpy
• pillow
• feather-format

Usage

1. Installation

$ git clone https://github.com/imoken1122/GIST-feature-extractor.git

2. Use GIST-feature extractor

Change these parameters if necessary.

param = {
        "orientationsPerScale":np.array([8,8]),
         "numberBlocks":[10,10],
        "fc_prefilt":10,
        "boundaryExtension": 10
        
}

Specifies an image name or a folder path containing several images and output path saving extracted gist-feature as option (Extention of output file is ".feather")

--input_path <your path>
--output_path <your path>

Let's extracting GIST-feature !! The following is an example.

$ python main.py --input_path image_name.png --output_path gist.feather

or

$ python main.py --input_path image_list --output_path gist.feather