1.use labelimg software for drawing bbox on top of images
2.select XML/PascalVOC format for doing annotation
3.Seprate the output in different folder ex. train==>> all images ; trainannotaion==>> xml format
4.we can not use XML/VOC format for object detecton in detectron2 , so we are converting xml to json using this file
python voc2coco.py train_annotations/ output.json
folder strcture==>> images -all images ; trainval.json- single json with all labels
python labelme2coco.py foldername --output trainval.json