Discover published sets by community

Purpose: To detect and recognize text within images for OCR (Optical Character Recognition). Annotation Process: Label text regions and transcribe the text within the bounding box or polygon surrounding the text.

Purpose: To annotate objects with more complex shapes with higher precision than bounding boxes. Annotation Process: Annotators draw polygons with multiple points around the contours of the object.

Purpose: To classify each pixel in the image into a corresponding class of object. Annotation Process: Label every pixel in the image so that pixels which are part of the same object category receive the same label.

Purpose: To assign a single label to an entire image, classifying the content as a whole. Annotation Process: Annotators view the image and assign one or multiple labels that best describe the entire image.

Purpose: To map out paths and borders within images, such as roads, edges, or product defects. Annotation Process: Draw lines that represent the linear features of the abovementioned objects or elements within an image.

Purpose: To identify specific points of interest in the object, which are crucial for understanding poses or gestures. Annotation Process: Mark specific predefined points (keypoints) on an object, such as the joints on a human body.

Purpose: To identify each instance of every object class within an image. Annotation Process: Similar to semantic segmentation, but in addition, each instance of an object is uniquely identified even if they belong to the same class.

Purpose: To give a three-dimensional perspective to objects for better depth understanding. Annotation Process: Draw 3D cuboids around objects, defining their length, width, and height in the 3D space.

Purpose: To identify and mark important features or objects within an image that act as reference points. Annotation Process: Place markers on significant features or areas in an image, which could be used for aligning or morphing images.