One Image Annotation Platform for All Your ML Needs

Used for finding the contour of an object like a silhouette or defining an amorphous entity like “sky” or “road".

Semantic/instance/volumetric segmentation

Semantic objects, like floor surfaces

Any visible object

Object detection

Composite objects

A lightweight vector, but drawn or erased like a painting tool. Handy for creating complex polygons and holes.

Composite objects, thin objects, or objects with holes

Semantic objects, like floor surfaces

Semantic/instance segmentation

Object detection

Used for 3D and 6 degree-o-freedom object detectors. Cuboids are 3-dimensional bounding boxes that can enclose an item, defining its dimension and pose in a captured scene.

Objects on flat planes that need to be navigated, such as cars

Objects that require robotic grasping

3D cuboid and 6DoF pose estimation

Object detection

An oval that appears as a two-diameter circle or multi-point polygon on export. Can be used as a round polygon or as an ellipse to mark circular objects that may distort with perspective.

Circular or oval objects

Items that can be regressed with 2-3 values (diameters, centroid)

Instance and semantic segmentation

Ellipsoid regression

A series of points forming a line that can be used for defining a slope, direction, or edge. This tool comes in handy for lane markings or trajectories.

Linearly defined objects with no volume, such as edges

Non-visual objects such as mid-points or trajectories

Regressor networks

Keypoint detectors

Bounding boxes are used for training detectors. They’re less precise than polygon masks but may prove cheaper to compute.

Uniformly shaped objects

Objects that don’t overlap

Low-compute projects

Defines a point that may represent an object or marker. Keypoints can be used individually or in groups to form a point map to define the pose of an object.

Object markers, such as face points or corners

Non-visual objects, such as the inner midpoint of an object

Keypoint detection and mapping

2D and 6DoF object pose

A network of keypoints connected by vectors. Used for defining the 2D or 3D pose of a multi-limbed object. Keypoint skeletons use a defined, moveable set of points that adapt to the object’s appearance.

Human, animal, or object pose

Objects with a defined number of marker points

Pose estimation

Tags describe the whole image for classification purposes. Unlike other annotations, they don’t apply to an image area. They describe features of an image as a whole, or define an image category.

Visual features that are present but not positional, such as “overexposed”

Objects taking up 50% or more than an image

Image categories, such as “indoor”

Image or video classifiers

Attributes refer to tags on an annotation. They describe objects in greater detail than the class name itself. They can define discrete or continuous features, such as color or age. Attributes help AI classify objects after detecting them.

Any object with relevant variable traits

Any model that can support an attribute-defining head

Instance IDs (also referred to as “object IDs” or “tracking IDs”) let you re-identify a specific object throughout a dataset. Each annotation is given a numeric ID so the object may be tracked or distinguished from other objects of the same class in an image.

Datasets where tracking is important

Instance/3D volumetric segmentation

Video object tracking

Text can be used to train OCR or to include unique textual information about an annotation. Unlike attributes, the text isn’t saved to be re-applied to other annotations.

Entering freeform text, such as in clinical notes or transcribed text

Optical Character Recognition (OCR)

Text detection

Defines a value between 0 and 360 to indicate the 2D pose or direction of an object. These vectors can be used to predict movement or define an object’s tilt.

Solid objects with a relevant direction or pose

Object Direction or 2D object pose

When your team's needs go beyond the standard tools, you can write plugins to add new functionalities and annotation types, giving you the freedom to expand the platform.

placeholder

placeholder