AiliaDetector Class Reference

Classes
class	AILIADetectorObject

Public Member Functions
static int	ailiaCreateDetector (ref IntPtr detector, IntPtr net, UInt32 format, UInt32 channel, UInt32 range, UInt32 algorithm, UInt32 category_count, UInt32 flags)
static void	ailiaDestroyDetector (IntPtr detector)
	Destroy the detector object. More...
static int	ailiaDetectorCompute (IntPtr detector, IntPtr src, UInt32 src_stride, UInt32 src_width, UInt32 src_height, UInt32 src_format, float threshold, float iou)
	Perform object recognition. More...
static int	ailiaDetectorGetObjectCount (IntPtr detector, ref UInt32 obj_count)
	Get the number of recognition results. More...
static int	ailiaDetectorGetObject (IntPtr detector, [In, Out] AILIADetectorObject obj, UInt32 obj_idx, UInt32 version)
	Get the recognition result. More...
static int	ailiaDetectorSetAnchors (IntPtr detector, float [] anchors, UInt32 anchors_count)
	Set anchors or biases information for YoloV2 etc. More...
static int	ailiaDetectorSetInputShape (IntPtr detector, UInt32 input_width, UInt32 input_height)
	Specifies the input image size to the model in YoloV3. More...

Public Attributes
const Int32	AILIA_DETECTOR_OBJECT_VERSION = (1)
const Int32	AILIA_DETECTOR_ALGORITHM_YOLOV1 = (0)
const Int32	AILIA_DETECTOR_ALGORITHM_YOLOV2 = (1)
const Int32	AILIA_DETECTOR_ALGORITHM_YOLOV3 = (2)
const Int32	AILIA_DETECTOR_ALGORITHM_YOLOV4 = (3)
const Int32	AILIA_DETECTOR_ALGORITHM_YOLOX = (4)
const Int32	AILIA_DETECTOR_ALGORITHM_SSD = (8)
const Int32	AILIA_DETECTOR_FLAG_NORMAL = (0)

Member Function Documentation

ailiaCreateDetector()

static int AiliaDetector.ailiaCreateDetector	(	ref IntPtr	detector,
		IntPtr	net,
		UInt32	format,
		UInt32	channel,
		UInt32	range,
		UInt32	algorithm,
		UInt32	category_count,
		UInt32	flags
	)

ailiaDestroyDetector()

static void AiliaDetector.ailiaDestroyDetector

(

IntPtr

detector

)

Destroy the detector object.

Parameters

detector

Detection object pointer

ailiaDetectorCompute()

static int AiliaDetector.ailiaDetectorCompute	(	IntPtr	detector,
		IntPtr	src,
		UInt32	src_stride,
		UInt32	src_width,
		UInt32	src_height,
		UInt32	src_format,
		float	threshold,
		float	iou
	)

Perform object recognition.

Parameters

detector	Detection object pointer
src	Image data (32bpp)
src_stride	Byte count of one line
src_width	Image width
src_height	Image height
src_format	Image format (AILIA_IMAGE_FORMAT_*)
threshold	Detection threshold (e.g. 0.1f) (the smaller the threshold, the more likely it is to be detected and the more detections will be made)
iou	Duplicate exclusion threshold (e.g. 0.45f) (the smaller the threshold is, the less duplicates are allowed and the fewer the number of detections)

Returns

If successful, return AILIA_STATUS_SUCCESS, otherwise return error code.

ailiaDetectorGetObject()

static int AiliaDetector.ailiaDetectorGetObject	(	IntPtr	detector,
		[In, Out] AILIADetectorObject	obj,
		UInt32	obj_idx,
		UInt32	version
	)

Get the recognition result.

Parameters

detector	Detection object pointer
obj	object information
obj_idx	object index
version	AILIA_DETECTOR_OBJECT_VERSION

Returns

Returns AILIA_STATUS_SUCCESS if successful, otherwise returns error code.

If ailiaPredict() has never been executed, then return * AILIA_STATUS_INVALID_STATE. Recognition results are sorted in order of probability.

ailiaDetectorGetObjectCount()

static int AiliaDetector.ailiaDetectorGetObjectCount	(	IntPtr	detector,
		ref UInt32	obj_count
	)

Get the number of recognition results.

Parameters

detector	Detection object pointer
obj_count	Number of objects

Returns

If successful, return AILIA_STATUS_SUCCESS, otherwise return error code.

ailiaDetectorSetAnchors()

static int AiliaDetector.ailiaDetectorSetAnchors	(	IntPtr	detector,
		float []	anchors,
		UInt32	anchors_count
	)

Set anchors or biases information for YoloV2 etc.

Parameters

detector	Detection object pointer
anchors	Dimensions of anchors (possible shape, height and width of detection box)
anchors_count	number of anchors (half of the array size of anchors)

Returns

Return AILIA_STATUS_SUCCESS if successful, otherwise return error code.

YoloV2, for example, tries multiple forms of the default detection box at the same time. This data describes the plural form of the box. anchors is stored in the form {x,y,x,y...} The anchors are stored in the form {x,y,x,y...}. If anchors_count is 5, then anchors is a 10-dimensional array.

ailiaDetectorSetInputShape()

static int AiliaDetector.ailiaDetectorSetInputShape	(	IntPtr	detector,
		UInt32	input_width,
		UInt32	input_height
	)

Specifies the input image size to the model in YoloV3.

Parameters

detector	Detection object pointer
input_width	Input image width for the model
input_height	Input image height of the model

Returns

Returns AILIA_STATUS_SUCCESS if successful, otherwise returns an error code.

YoloV3 allows a single model to correspond to any input resolution. (with a multiple limit of 32) Execute this API if you want to specify the input image size for a model, e.g. to reduce computation time. This API must be executed between ailiaCreateDetector() and ailiaDetectorCompute(). If this API is not executed, the default value of 416x416 is used. If executed outside of YOLOv3, it returns AILIA_STATUS_INVALID_STATE.

Member Data Documentation

AILIA_DETECTOR_ALGORITHM_SSD

const Int32 AiliaDetector.AILIA_DETECTOR_ALGORITHM_SSD = (8)

SSD

AILIA_DETECTOR_ALGORITHM_YOLOV1

const Int32 AiliaDetector.AILIA_DETECTOR_ALGORITHM_YOLOV1 = (0)

YOLOV1

AILIA_DETECTOR_ALGORITHM_YOLOV2

const Int32 AiliaDetector.AILIA_DETECTOR_ALGORITHM_YOLOV2 = (1)

YOLOV2

AILIA_DETECTOR_ALGORITHM_YOLOV3

const Int32 AiliaDetector.AILIA_DETECTOR_ALGORITHM_YOLOV3 = (2)

YOLOV3

AILIA_DETECTOR_ALGORITHM_YOLOV4

const Int32 AiliaDetector.AILIA_DETECTOR_ALGORITHM_YOLOV4 = (3)

YOLOV4

AILIA_DETECTOR_ALGORITHM_YOLOX

const Int32 AiliaDetector.AILIA_DETECTOR_ALGORITHM_YOLOX = (4)

YOLOX

AILIA_DETECTOR_FLAG_NORMAL

const Int32 AiliaDetector.AILIA_DETECTOR_FLAG_NORMAL = (0)

No options

AILIA_DETECTOR_OBJECT_VERSION

const Int32 AiliaDetector.AILIA_DETECTOR_OBJECT_VERSION = (1)

---

The documentation for this class was generated from the following file:

• AiliaDetector.cs