[Question]It takes a long time to change to coco json type.

[udkii]

Hi, wkentaro. I have some problem to use 'labelme2coco.py '

After labeling instance segmentation with labelme, it takes a long time to convert to coco json type.

I need to change the images labeled using labelme to coco type. There are a total of 229 images in my image, and 458 images including the json file. And it takes over an hour to convert all these images to coco type using labelme2coco.py .

1. Is there any way to speed up the conversion to coco files?
2. Every time a new image is added to the train dataset, the entire image is changed again using labelme2coco.py. Is there a more efficient way? (For example, is there a way to leave the previously converted files as they are and convert only the new images using labelme2coco.py and then combine them?)

Thank you so much for reading.