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
• Run pip install -r requirements.txt to install needed libs

How to use Git

• Clone git to your machine

  • Start a terminal (powershell on Windows)
  • Switch to the directory you'd like to clone to
  • Run command git clone https://github.com/LiuTed/FontRecognition.git

• Pull code from git

  • Run command git pull

• Push your change to git

  • Run command git add <files you changed>
  • Run command git commit -m 'your description to the changes'
  • Run command git push
  • If push failed and git gives the warning that you need to pull first, run git pull

• How to merge

  • Merge may happen when you pull from git
  • Find the files it lists
  • You would find part of the file looks like

  
    >>> 1234567890abcdef  
    code1
    ----------------------------------
    code2
    <<< fedcba0987654321
    

  • There are some conflict between code1 and code2, you need to merge them manually.
    Replace this part (the two hex number included) with the final code
  • Run git commit and save the description it automatically generated
  • Run git push

Jobs

• Proposal
  • Done
• Preprocessing: Read from csv file and image reconstruction
  • Done
• Feature Extraction: Extract feature from image and (optional) encoding
  • SIFT: Liu
  • Fisher Vector: Liu
• Model implementation & training
  • Bayesian Inference:
  • Multiclass SVM:
  • Decision Tree (Random Forest):
  • Clustering:
  • ResNet-18 & LeNet: