VIAtoYOLO is a tool to convert labeled data from VGG Image Annotator (VIA) tool to YOLO format to train instance segmentation algorithms on YOLO series like YOLOv9. This is an implementation on Python 3.
- First of all, you need to annotate all the images you will use as training, validation and testing data. To label the features to be detected in them use VIA tool from the University of Oxford (Dutta and Zisserman, 2019).
VIA: https://www.robots.ox.ac.uk/~vgg/software/via/via_demo.html
YOLO segmentation format for one class: 0 x1 y1 x2 y2 x3 y3 x4 y4 ... xn yn
- Save all the labeled JSON files you want to convert to YOLO format in the
/VIA/folder. Now run the python codeviatoyolo. Ensure you indicate the size of the images in pixels both horizontally and vertically--img 1024 768.
python3 viatoyolo.py --img 1024 768- The resulting labeled data converted to yolo format (for one class) is saved in
/yolo/folder. You will find a TXT file for each image inside the JSON files.
YOLOv9: https://github.com/WongKinYiu/yolov9/
To cite this repository:
Berganzo-Besga, I. VIAtoYOLO: YOLO labeling tool for instance segmentation models. GitHub repository 2024. Available online: https://github.com/iberganzo/VIAtoYOLO