Arabic-Caligraphy-Font-Identification

An implementation of a classifier that distinguishes between 9 different arabic fonts.

This is CMPN450 course project - Faculty of Engineering, Cairo University.

Implementation

Local Phase Quantaization (LPQ) feature is calculated for each test image and used to train a Support Vector Machine (SVM).

To use this code

To install needed packages use:

pip install -r requirements.txt

To run the trained model on testcases in /test use:

python predict.py

To calculate the accuracy use:

python evaluate.py

Input Files:

/test

This is the directory where the test images are located. The output is ordered ascendingly according to the file names.

ground_truth.txt

Contains the actual labels of the test images, each in a seperate line.

Output Files (/out):

/out/results.txt

Contains the class of each test image in ascending order according to the image name, each in a seperate line.

/out/time.txt

Contains the classification time of each test image in ascending order according to the image name, each in a seperate line.