This project consist of 3 main file corresponding with specific function and some helper folder
File
- myanmar_handwriting_text_generation.py (used to generate
myanmar handwriting text
image) - custom_text_generation_v2.py (generate
vietnamese text image
with custom erode operation) - text_generation.py (generate random text)
Directory:
- background_cccd_new (background for the cccd)
- background (background for cccd chip)
- data (data to generate custom text)
- fonts (font for text)
- myanmar_project/crop_bg_image (background for myanmar)
- myanmar_project/crop_data (myanmar cropped text image)
Following below step to run
Step 1: Prepare
-
myanmar_label.txt file
-
myanmar images in myanmar_project/crop_data
-
myanmar background in myanmar_project/crop_bg_image
Step 2: Run
python myanmar_handwriting_text_generation.py
The output image and label will appear in your save_image_path
and save_path
The file generate text image following 5 step:
-
Choice
text box coordinate
based on type card (cccd chip/the cccd) and type text(hokhauthuongtru/nguyenquan) -
Generate
text image and mask of text
with PIL (text with black and background with white color) -
Blend text with background by random weight -> make text more natural
-
Paste only black text into embryo background by using
mask of text
and text box coordinate -
Crop text from embryo background
-
Blur only edge of text -> make text more thin but keep the text sketon
Following below step to run
Step 1: Prepare new background (the cccd) in background_cccd_new and former background (cccd_chip) in background
Step 2: Run this file following below instructions
This file have 2 mode:
- generate custom text (use
string
argument). Run this file by following command:
python custom_text_generation_v3.py
- generate exist text from
.txt
file. Run this file by following command:
python custom_text_generation_v3.py --text_from_file <text file>
Note: This file only support hokhauthuongtru
and nguyenquan
field
This file is used for generate .txt file as input of custom_text_generation_v2.py file. You can simply run this by:
python text_generation.py
a synthetic_data.txt
file will be created