Landmark-Annotator

Prerequisites

Python3

• If you already have downloaded python3 before, you do not have to download it again
• If you do not have python3 installed, go to the link below and download python3 https://www.python.org/downloads/

Download libraries

• Type the following in your terminal

pip3 install opencv-python argparse

PNG files

If you are dealing with dicom files, you need to convert the files into PNG format. Please refer to https://github.com/yehyunsuh/DICOM-to-PNG.

1. When you have PNG files in png folder

1.1 How the folder should look like

Dicom
├─ png
└─ landmark_annotator.py

1.2 Run python file

• Type the following in your terminal

cd <<path to Landmark-Annotator Folder>>
python3 landmark_annotator.py

2. When you have PNG files in other folder

2.1 How the folder should look like

Dicom
├─ <<name of the png folder>>
└─ landmark_annotator.py

2.2 Run python file

• Type the following in your terminal

cd <<path to Landmark-Annotator Folder>>
python3 landmark_annotator.py --path <<name of the png folder>>

For example, if your folder name is human_data,

cd <<path to Landmark-Annotator Folder>>
python3 landmark_annotator.py --path human_data

3. How to annotate the markers

• left click: every time you do a click, there will be a red dot generated in the image and coordinates of the red dot will be extracted

• b: when you annotate the wrong point, press b and the dot will be erased

• n: when you are done with one image, press n and you can move on to the next image

• q: when you are done with annotating, press q and program will be terminated

• After executing the file, you will have a txt file that has current date + current time + folder name.txt

Update on functions (2023.12.09)

• checkpoint: your checkpoint will be created in checkpoint folder in the name of your path
  • Even if you restart the program, if you have already done an annoation, this program will skip the image
  • If you want to re-annotate an image that you have already annotated without using the p fuction, just erase the name of the image in the checkpoint/<<name of path>>.txt file
• p: when you want to go to previous image, press p and you can move to the previous image.
  • If you go back to a specific image and re-annotate and move to the next images, the previous annotation will be overlapped with the new annotation
  • If you do not do any annotations, it will just skip the image without overlapping any annotations

Acknowledgement

Most of the initial stage code has used example from https://gaussian37.github.io/vision-opencv-coordinate_extraction/.