ChartClassifer

An chart image classifier based on Python

Copyright (c) 2013 Yi Li [email protected]

INSTALL

Prerequisites

Mandatory

• Python (2.7). Python 2.7.3 is recommended.

• Numpy(>=1.6.1). You can download the source of Numpy from here.

• Scipy(>=0.10). You can download the source of Scipy from here.

• PIL(>=1.1.7).

Install numpy, scipy on Ubuntu: sudo apt-get install python-numpy python-scipy

Instal PIL on Linux:

1. Download PIL 1.1.7 under "Python Imaging Library 1.1.7 Source Kit" as Imaging-1.1.7.tar.gz

2. run: tar -xzvf Imaging-1.1.7.tar.gz

3. Go into the folder created and (as root) install the package as run: python setup.py install

Compile libSVM

Under the bin directory, imageUtils.py and img_svm_predict.py are my sources; libSVM_COPYRIGHT Makefile svm.cpp svm.def svm.h svm.py svmutil.py are the sources of libSVM.

The static library of libSVM need to be compiled to run img_svm_predict.py. Simply type:

make

TEST

Under the model directory are the trained model of chart image classifier. Under the images directory are the images used for both training and testing.

To test img_svm_predict.py, go to images directory and run:

python ../bin/img_svm_predict.py ./bar/1.jpeg

python ../bin/img_svm_predict.py ./pie/2.jpeg

You can also download any chart images from Google image to test. Currently, only bar chart, pie chart and scatter plot are supported.

Reference

Manolis Savva (2011). ReVision: Automated Classification, Analysis and Redesign of Chart Images