Landmark GUI

Simple interface for manually annotating face image data (68 landmarks).

Preview

Controls:

• Left Arrow / Right Arrow : Previous image / Next image
• CTRL+S : Save current landmark to file (overwrite)
• LeftClick : Drag a point
• CTRL+LeftClick : Drag a group

Data setup

Image and landmark data should exist in a directory like so:

 your_data
 |
 ├─── img_0.png
 ├─── img_0_ldmks.txt
 ├─── img_1.png
 ├─── img_1_ldmks.txt
 ├─── img_2.png
 ├─── img_2_ldmks.txt
 │    ...

img_i_ldmks.txt encodes the landmark information as 68 coordinate pairs:

136.586700 181.368881
139.143250 211.010727
143.046860 235.316269
215.864960 347.593414
...
266.220978 347.455549

Install

sudo chmod +x setup.sh
./setup.sh

Usage

Make sure python environment is activated

source .venv/bin/activate

Launch GUI app

python3 main.py --dir <your_image_dir_path>

Open a data directory:

List of controls:

• Left Arrow / Right Arrow : Previous image / Next image
• CTRL+S : Save current landmark to file (overwrite)
• LeftClick : Drag a point
• CTRL+LeftClick : Drag a group
• CTRL+D : Delete image and landmark data, go to Next image