Index

Setup

Recording Single Poses

Detecting Single Poses

Deleting and Editing Single Poses

Setup

To start setting up the Single Pose Recognition System, A data asset must be created.

The data asset must be of type SinglePosesDataAsset. Any other data asset type will not be accepted by the system. The poses which will be recorded and detected will be saved in the data structure. There is no need to add elements to the data asset since the system will add poses through recording.

The single pose recognition is an actor component that can be added to a character or a pawn. To start setting up a pose recognition character add the actor components to an actor, pawn or character.

In the single pose recognition component, under Single Pose Recognition Data Asset, the right data asset must be assigned to the corresponding data asset slot.

Important: The body area which will be used must have its own data asset. Not assigning a data asset to a body part will disable recording for the body part.

Important: A pose recognition character or pawn must contain a skeletal mesh component.

In both the static and dynamic pose recognition components, under Skeleton Setup, the bone name must be assigned to the corresponding array element.

Lastly, in the EventBegin event, the InitMesh function must be called with the Skeletal Mesh as a parameter.

Recording Single Poses

To record a single simple pose, one of the record single pose functions must be called. The single pose system includes pose recording for full body, left hand, right hand, upper body and lower body.

Tip: The functionality of detecting and recording should not run at the same time. To ensure that does not happen a boolean variable can be made to decide if the system is detecting or recording poses.

The single pose record functions include a pose name and a pose type. The function should be connected to an input action or a button press. When the button is pressed a new pose will be recorded. The pose name does not need to be unique when recording as the system will create a new entry in the data asset. The name should be changed before starting detecting poses so that every pose has a unique name.

Detecting Single Poses

The first step in detecting single poses is binding events. There is one event per body area when detecting single poses: pose performed. The event will return a string with the current pose name.

The system gives access to the current body area pose type.

After binding to the event, the StartDetect_BodyArea_Poses function must be called once (E.g., in Event BeginPlay or performing an input action).

In the single pose recognition details panel, under single pose recognition detecting variables, a threshold and a timer variable can be set. The time variable is in seconds. The threshold should have a value between 0 and 1. The timer refers to the time between detecting the poses (similar to EventTick).

Deleting and Editing Single Poses

Deleting and editing can be done within the data asset. The pose name and type can be edited. The bone transforms cannot be edited.