Skip to content

Add action and gesture recognition #373

Description

@alexarje

We should implement a form of action and gesture recognition to the toolbox. From our terminology, motion is the continuous displacement in space over time. Actions are motion segments, often (but not always) with a clear beginning and end. Actions can be goal-directed but not always (eg in dance). In music, Actions are often sound-producing.

Gestures are actions with some kind of meaning. The meaning can be direct and connected to natural language, but the meaning can also be expressive, such as in music and dance.

We should implement a multilayered approach in MGT, where actions are derived from motion, and gesture is added to actions (not all).

It is important that action and gesture recognition is based on human activity in a video, not from camera motion or scene change.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions