Detailed Description

Detect regions in the view which are blocked.

Note: SW Release Applicability: These APIs are available in NVIDIA DRIVE Software releases.

Data Structures
struct	dwBlindnessDetectionOutput
	Holds the outputs for ClearSightNet. More...

struct	dwBlindnessDetectorParams
	Holds init params for Blindness Detector. More...

Macros
#define	DW_CLEARSIGHTNET_DETECTOR_MAX_IMAGES 16
	Maximum number of images to run detector on. More...

#define	DW_CLEARSIGHTNET_NUM_MAX_REGIONS 8
	Number of sub-regions in each direction. More...

#define	DW_CLEARSIGHTNET_NUM_OUTPUT_CHANNELS 3
	Number of ClearSightNet output channels. More...

Typedefs
typedef struct dwBlindnessDetectorObject *	dwBlindnessDetectorHandle_t
	Handle to a BlindnessDetector. More...

Functions
DW_API_PUBLIC dwStatus	dwBlindnessDetector_detect (const dwImageCUDA *inputImage, dwBlindnessDetectorHandle_t handle)
	Runs asynchronous inference using the provided DNN model. More...

DW_API_PUBLIC dwStatus	dwBlindnessDetector_getCUDAStream (cudaStream_t *stream, dwBlindnessDetectorHandle_t handle)
	Gets CUDA stream used by the Blindness detection. More...

DW_API_PUBLIC dwStatus	dwBlindnessDetector_getOutput (dwBlindnessDetectionOutput *output, dwBlindnessDetectorHandle_t handle)
	Return a blindness detection output. More...

DW_API_PUBLIC dwStatus	dwBlindnessDetector_initDefaultParams (dwBlindnessDetectorParams *params)
	Initializes default parameters for Blindness detector module. More...

DW_API_PUBLIC dwStatus	dwBlindnessDetector_initialize (dwBlindnessDetectorHandle_t handle, const dwBlindnessDetectorParams params, dwContextHandle_t ctx)
	Initializes a Blindness detector module. More...

DW_API_PUBLIC dwStatus	dwBlindnessDetector_release (dwBlindnessDetectorHandle_t handle)
	Releases the Blindness detector module. More...

DW_API_PUBLIC dwStatus	dwBlindnessDetector_reset (dwBlindnessDetectorHandle_t handle)
	Resets the Blindness detector module. More...

DW_API_PUBLIC dwStatus	dwBlindnessDetector_setCUDAStream (cudaStream_t stream, dwBlindnessDetectorHandle_t handle)
	Sets the CUDA stream for CUDA related operations. More...

Data Structure Documentation

◆ dwBlindnessDetectionOutput

struct dwBlindnessDetectionOutput

Data Fields
float32_t	blindnessRatio	Overall blindness ratio value (between 0 and 1)
float32_t	fullBlindRatio	Full blindness ratio value (between 0 and 1).
dwImageHandle_t	mask	RGBA mask (of same size as output blob) with multiple class information encoded into different channels.
uint8_t	numRegionsX	number of region dividers in X
uint8_t	numRegionsY	number of region dividers in Y
float32_t	partBlindRatio	Partial blindness ratio value (between 0 and 1).
float32_t	regionBlindnessRatio[DW_CLEARSIGHTNET_NUM_MAX_REGIONS][DW_CLEARSIGHTNET_NUM_MAX_REGIONS]	overall blindness ratios for each sub-region
float32_t	regionDividersX[DW_CLEARSIGHTNET_NUM_MAX_REGIONS - 1]	region dividers in X (in output blob pixels)
float32_t	regionDividersY[DW_CLEARSIGHTNET_NUM_MAX_REGIONS - 1]	region dividers in Y (in output blob pixels)
float32_t	regionFullBlindRatio[DW_CLEARSIGHTNET_NUM_MAX_REGIONS][DW_CLEARSIGHTNET_NUM_MAX_REGIONS]	full blindness ratios for each sub-region
float32_t	regionPartBlindRatio[DW_CLEARSIGHTNET_NUM_MAX_REGIONS][DW_CLEARSIGHTNET_NUM_MAX_REGIONS]	partial blindness ratios for each sub-region

◆ dwBlindnessDetectorParams

struct dwBlindnessDetectorParams

Data Fields
dwClearSightNetHandle_t	clearSightNetHandle	handle to ClearSightNet DNN module
uint8_t	numRegionsX	number of region dividers in X
uint8_t	numRegionsY	number of region dividers in Y
float32_t	regionDividersX[DW_CLEARSIGHTNET_NUM_MAX_REGIONS - 1]	region dividers in X (in image fractions)
float32_t	regionDividersY[DW_CLEARSIGHTNET_NUM_MAX_REGIONS - 1]	region dividers in Y (in image fractions)
uint32_t	temporalFilterWindow	num frames over which to temporally filter blindness ratio

Macro Definition Documentation

◆ DW_CLEARSIGHTNET_DETECTOR_MAX_IMAGES

#define DW_CLEARSIGHTNET_DETECTOR_MAX_IMAGES 16

Maximum number of images to run detector on.

Definition at line 63 of file BlindnessDetector.h.

◆ DW_CLEARSIGHTNET_NUM_MAX_REGIONS

#define DW_CLEARSIGHTNET_NUM_MAX_REGIONS 8

Number of sub-regions in each direction.

Definition at line 69 of file BlindnessDetector.h.

◆ DW_CLEARSIGHTNET_NUM_OUTPUT_CHANNELS

#define DW_CLEARSIGHTNET_NUM_OUTPUT_CHANNELS 3

Number of ClearSightNet output channels.

Definition at line 66 of file BlindnessDetector.h.

Typedef Documentation

◆ dwBlindnessDetectorHandle_t

typedef struct dwBlindnessDetectorObject* dwBlindnessDetectorHandle_t

Handle to a BlindnessDetector.

Definition at line 108 of file BlindnessDetector.h.

Function Documentation

◆ dwBlindnessDetector_detect()

DW_API_PUBLIC dwStatus dwBlindnessDetector_detect	(	const dwImageCUDA *	inputImage,
		dwBlindnessDetectorHandle_t	handle
	)

Runs asynchronous inference using the provided DNN model.

The inference pipeline is as follows:

DataPreparation: Input images are prepared by scaling, padding and other transformations set in dataConditionerParams at initialization time in order to have the right properties required by the DNN. If the given DNN model expects batched input, the given images are divided into batches, where each batch has the expected batch size.
Inference: Inference is run on the prepared image in an asynchronous fashion. This means that an inference call can be queued using dwBlindnessDetector_detect() and then other computation can be performed and dwBlindnessDetector_getOutput(), whenever called, will automatically wait for inference to finish before returning the output.

Parameters

[in]	inputImage	Image where the detector is to be applied.
[in]	handle	Specifies the BlindnessDetector handle.

Returns: DW_SUCCESS
DW_INVALID_HANDLE -If the given context handle is invalid,i.e. null or of wrong type
DW_BAD_CAST
DW_INVALID_ARGUMENT

Deprecated:: Will be removed. please consider using BlindnessDetector_processPipeline apis dwBlindnessDetector_bindInput(), dwBlindnessDetector_bindOutput(), dwBlindnessDetector_process() or dwBlindnessDetector_processDLA()

◆ dwBlindnessDetector_getCUDAStream()

DW_API_PUBLIC dwStatus dwBlindnessDetector_getCUDAStream	(	cudaStream_t *	stream,
		dwBlindnessDetectorHandle_t	handle
	)

Gets CUDA stream used by the Blindness detection.

Parameters

[out]	stream	The CUDA stream currently used.
[in]	handle	A handle to the BlindnessDetector module.

Returns: DW_INVALID_HANDLE - If the given context handle is invalid,i.e. null or of wrong type
DW_SUCCESS

◆ dwBlindnessDetector_getOutput()

DW_API_PUBLIC dwStatus dwBlindnessDetector_getOutput	(	dwBlindnessDetectionOutput *	output,
		dwBlindnessDetectorHandle_t	handle
	)

Return a blindness detection output.

Parameters

[out]	output	Struct encoding blindness detection output
[in]	handle	A BlindnessDetector handle

Returns: DW_INVALID_HANDLE - If the given context handle is invalid,i.e. null or of wrong type
DW_BAD_CAST - If cannot cast the given handle to the expected type.
DW_SUCCESS

Deprecated:: Will be removed. please consider using BlindnessDetector_processPipeline apis dwBlindnessDetector_bindInput(), dwBlindnessDetector_bindOutput(), dwBlindnessDetector_process() or dwBlindnessDetector_processDLA()

◆ dwBlindnessDetector_initDefaultParams()

DW_API_PUBLIC dwStatus dwBlindnessDetector_initDefaultParams ( dwBlindnessDetectorParams * params )

Initializes default parameters for Blindness detector module.

Parameters

[out] params BlindnessDetector parameters.

Returns: DW_SUCCESS, DW_INVALID_ARGUMENT

Note: Default precision depends on the current GPU.

◆ dwBlindnessDetector_initialize()

DW_API_PUBLIC dwStatus dwBlindnessDetector_initialize	(	dwBlindnessDetectorHandle_t *	handle,
		const dwBlindnessDetectorParams *	params,
		dwContextHandle_t	ctx
	)

Initializes a Blindness detector module.

Parameters

[out]	handle	A pointer to the Blindness detection algorithm.
[in]	params	Struct encoding config parameters.
[in]	ctx	Specifies the handler to the context to create BlindnessNet.

Returns: DW_INVALID_ARGUMENT, DW_SUCCESS

◆ dwBlindnessDetector_release()

DW_API_PUBLIC dwStatus dwBlindnessDetector_release ( dwBlindnessDetectorHandle_t handle )

Releases the Blindness detector module.

Parameters

[in] handle The object handle to release.

Returns: DW_INVALID_HANDLE - If the given context handle is invalid,i.e. null or of wrong type
DW_BAD_CAST - If cannot cast the given handle to the expected type.
DW_SUCCESS

Note: This method renders the handle unusable.

◆ dwBlindnessDetector_reset()

DW_API_PUBLIC dwStatus dwBlindnessDetector_reset ( dwBlindnessDetectorHandle_t handle )

Resets the Blindness detector module.

Parameters

[in] handle Specifies the detector to reset.

Returns: DW_INVALID_HANDLE - If the given context handle is invalid, i.e. null or of wrong type
DW_BAD_CAST - If cannot cast the given handle to the expected type.
DW_SUCCESS

◆ dwBlindnessDetector_setCUDAStream()

DW_API_PUBLIC dwStatus dwBlindnessDetector_setCUDAStream	(	cudaStream_t	stream,
		dwBlindnessDetectorHandle_t	handle
	)

Sets the CUDA stream for CUDA related operations.

Note: The ownership of the stream remains by the callee.

Parameters

[in]	stream	The CUDA stream to be used. Default is the one passed during initialization.
[in]	handle	A handle to the BlindnessDetector module for which to set CUDA stream.

Returns: DW_INVALID_HANDLE - If the given context handle is invalid,i.e. null or of wrong type
DW_SUCCESS

Detailed Description

Data Structures

Macros

Typedefs

Functions

Data Structure Documentation

◆ dwBlindnessDetectionOutput

◆ dwBlindnessDetectorParams

Macro Definition Documentation

◆ DW_CLEARSIGHTNET_DETECTOR_MAX_IMAGES

◆ DW_CLEARSIGHTNET_NUM_MAX_REGIONS

◆ DW_CLEARSIGHTNET_NUM_OUTPUT_CHANNELS

Typedef Documentation

◆ dwBlindnessDetectorHandle_t

Function Documentation

◆ dwBlindnessDetector_detect()

◆ dwBlindnessDetector_getCUDAStream()

◆ dwBlindnessDetector_getOutput()

◆ dwBlindnessDetector_initDefaultParams()

◆ dwBlindnessDetector_initialize()

◆ dwBlindnessDetector_release()

◆ dwBlindnessDetector_reset()

◆ dwBlindnessDetector_setCUDAStream()