Semi-Automatic Yolo Annotation Tool In Python
PyYAT |
Using this tool, we can annotate bounding boxes for image annotation in YOLO format. It uses YOLOv3-608 weights to pre-annotate the bounding boxes in images. It reduces time and efforts in annotating large datasets by upto 90%.
- 'yolo_annotation_tool.py' : Main file to load images one-by-one from the dataset, and then annotate them.
- 'recognize_objects.py' : Object recognition class for pre-annotating images before manual annotation process.
- 'config.ini' : Edit data folder, output folder and label file path according to your preference. 'annotation_index' is automatically updated based on index of the last saved annotated image.
- 'labels.csv' : List of all the classes to be annotated.
- '/models' : It contains YOLOv3-608 weights (to be downloaded), cfg and coco.names files.
- '/data' : Sample of input images to be annotated
- '/output' : Sample of output files after annotation.
python yolo_annotation_tool.py
- 'S' : Save annotations
- 'L' : Change current class of labeling
- 'Esc' : Exit the code
Image 1 | Image 2 | Image 3 |
Image 1 | Image 2 | Image 3 |