Class CameraObjectDetectionSaver

Main component for object annotation in camera images. Manages objects and cameras, dataset properties etc.

Inheritance

System.Object

CameraObjectDetectionSaver

Namespace: Marus.ObjectAnnotation

Assembly: cs.temp.dll.dll

Syntax

public class CameraObjectDetectionSaver : MonoBehaviour

Fields

CameraViews

Holds list of cameras to annotate objects from.

Declaration

public List<Camera> CameraViews

Field Value

Type	Description
List<Camera>

DatasetFolder

Declaration

public string DatasetFolder

Field Value

Type	Description
System.String

Enable

Enable/Disable switch. Usefull for starting annotation during runtime.

Declaration

public bool Enable

Field Value

Type	Description
System.Boolean

GenerateTestSubset

If enabled, creates test subset besides train and validation.

Declaration

public bool GenerateTestSubset

Field Value

Type	Description
System.Boolean

ImageHeight

Image height It should correspond to display resolution in Game Mode.

Declaration

public int ImageHeight

Field Value

Type	Description
System.Int32

ImageWidth

Image width It should correspond to display resolution in Game Mode.

Declaration

public int ImageWidth

Field Value

Type	Description
System.Int32

MinimumObjectArea

Area threshold for object to be annotated. In pixes squared.

Declaration

public float MinimumObjectArea

Field Value

Type	Description
System.Single

MinVerticalPosition

MinVerticalPosition (eg. water surface y value) is used for ignoring objects, or parts of object that are under water (not visible).

Declaration

public float MinVerticalPosition

Field Value

Type	Description
System.Single

ObjectClasses

Declaration

public List<ClassObjects> ObjectClasses

Field Value

Type	Description
List<ClassObjects>

RaycastCheck

Use raycasting for more precise results.

Declaration

public bool RaycastCheck

Field Value

Type	Description
System.Boolean

SaveFrequencyHz

Annotation frequency in Hz.

Declaration

public float SaveFrequencyHz

Field Value

Type	Description
System.Single

TestSize

Declaration

public int TestSize

Field Value

Type	Description
System.Int32

TrainSize

Declaration

public int TrainSize

Field Value

Type	Description
System.Int32

ValSize

Declaration

public int ValSize

Field Value

Type	Description
System.Int32

VertexStep

Step used when iterating over object vertices. Bigger step means less precision but better performance.

Declaration

public int VertexStep

Field Value

Type	Description
System.Int32

Methods

GetBoundingBoxFromMesh(GameObject, Camera, Int32, Single, Boolean)

Calculates bounding box of object in camera view based on mesh vertices.

Declaration

public static Rect GetBoundingBoxFromMesh(GameObject gameObject, Camera CameraView, int VertexStep = 20, float VerticalPositionLimit = 0F, bool RaycastCheck = true)

Parameters

Type	Name	Description
GameObject	gameObject	Game object
Camera	CameraView	Camera object
System.Int32	VertexStep	Vertex Step
System.Single	VerticalPositionLimit	Vertical position limit. Vertices below will be ignored
System.Boolean	RaycastCheck	Use raycast checking for more precise bounding boxes.

Returns

Type	Description
Rect	Bounding box of object in camera.