It is available at https://github.com/heartexlabs/labelImg/blob/master/README.rst. It is an open source, lightweight, fast, easy-to-use, Python-based application with a short learning curve.
(https://github.com/wkentaro/labelme). It is similar to labelImg in terms of being a Python-based application, open source, fast, and easy to use. However, it supports other annotation types besides rectangle bounding boxes. Specifically, it can annotate polygons, rectangles, circles, lines, and points. Therefore, it can be used to label training images for object detection and semantic or instance segmentation tasks.
broswer-based web application
every json file is stored in the same folder with all the images
- The info section provides general descriptions of the dataset
- The licenses section (licenses) is a list of licesnses applied to the images
- The categories section (categories) is a list of the available categories(or class lables) and supercategories for this dataset
- The images section is a list of image elements, and each has information such as the id,width,height ,and file_name of the image
- The annotations section is a list of annotations and each annotation has information such as segmentation and a bounding box (bbox)
Pascal VOC (pattern analysis,statistical modeling and computational learning visual object classes )
it's an XML file per image
it's similar to the PASCAL VOC annotation style,but it's .txt instead or maybe .yaml file