Draw bounding boxes on original images based on yolo format annotation. It can help you checking the correctness of annotation and extract the images with wrong boxes. You can use makesense.ai to re-annotate them.
- Put all of your raw images in
./raw_images/
folder. - Put all of your annotation files(YOLO format txt) in
./labels/
folder. - Write your class information into classes.txt.
- Run
python draw_box.py
in your terminal.
Now you have the images with bounding boxes in the ./save_image/
folder.
- Put the images with incorrect boxes into
./wrong/
folder. - Clean up
./save_image/
folder. - Run
python get_origin_image.py
in your terminal.
Now you have the raw images which were annotated incorrectly in the ./save_image/
folder.
Thanks for 批量将yolo-v3检测结果在原图上画矩形框显示. I modified the code and enhanced its ease of use, added multi-label processing and labeling of category names, and provided a method to obtain the original image.