Font Recognition

Pattern Recognition Project
Font data can be downloaded at: http://archive.ics.uci.edu/ml/datasets/Character+Font+Images

How to prepare runtime environment

Install python3 with pip (3.5 suggested)
Run pip install -r requirements.txt to install needed libs
You may need to install tensorflow or tensorflow-gpu (depend on your environment) if want to run CNN.py.

Proposal: Together, Done
Preprocessing: Read from csv file and image reconstruction -Liu, Done
Feature Extraction: Extract feature from image and (optional) encoding
- HOG: Done
- SIFT: patent encumbered, replaced by Daisy Descriptor, Done
Model implementation & training
- Bayesian Inference: Li
- Multiclass SVM: Li
- Decision Tree (Random Forest): Chen
- K-Means: Chen
- KNN: Liu
- Mean-shift: Liu
- DCNN: Liu
Report: Together, Done

Put the font data at ../fonts, or modify the path in each python file to the correct path containing data
Different method may need different type of data, but you can call python utils.py to convert the csv files to binary files (default stored in ../data) which suitable for most methods
CNN, KNN, mean-shift, NB, SVM contains the code of methods
feature_PCA_DAISY, net_def, summary, utils are helper files

Name		Name	Last commit message	Last commit date
Latest commit History 15 Commits
.gitignore		.gitignore
CNN.py		CNN.py
KNN.py		KNN.py
NB.py		NB.py
Pattern_Recognition_Project_Proposal.pdf		Pattern_Recognition_Project_Proposal.pdf
Pattern_Recognition_Project_Report.pdf		Pattern_Recognition_Project_Report.pdf
Readme.md		Readme.md
SVM.py		SVM.py
correspondence.txt		correspondence.txt
dt_md_on_accuracy.py		dt_md_on_accuracy.py
dt_mid_on_accuracy.py		dt_mid_on_accuracy.py
feature_PCA_DAISY.py		feature_PCA_DAISY.py
mean-shift.py		mean-shift.py
net_def.py		net_def.py
performance_of_dt&rf.py		performance_of_dt&rf.py
readme.txt		readme.txt
requirements.txt		requirements.txt
rf_N_on_accuracy.py		rf_N_on_accuracy.py
summary.py		summary.py
utils.py		utils.py