Tool for image annotation in context of modern object detection

A modern approach to object detection in digital images is based on machine learning, which usually requires a significant amount of annotated data. The annotation process of images involves assigning the class and the location of the objects of interest which can be a time-consuming and tedious task. In this paper, a tool for object annotation in digital images is presented. The tool implements two interfaces: one for project management and one for object annotation in form of bounding boxes. The project management interface allows the users to define the name of the project, choose the storage location, specify the classes of objects of interest, select the project gallery and choose the structured computer format for storing the annotation files. The object annotation interface allows users to create annotations for objects of interest, visualize them, manage the created annotations, and save them in the working annotation dictionary. Users can also configure additional parameters for exporting annotation files to the disk. To further streamline the annotation process, the tool offers automatic detection of objects of interest using the YOLOv5 detector.