Detailed Description

Defines an API for the GStreamer NvDsPreProcess plugin.

Defines an API for the GStreamer NvDsPreProcess custom lib.

Defines an API for the GStreamer NvDsPreProcess custom lib implementation.

Data Structures
struct	CustomTransformParams
	custom transformation parameter for calling nvbufsurftransform api for scaling and converting the ROIs to the network resolutions to be used by custom lib. More...

struct	NvDsPreProcessTensorParams
	Holds model parameters for tensor preparation. More...

struct	NvDsPreProcessNetworkSize
	Holds information about the model network. More...

struct	CustomSeqProcTensorParams
	Tensor params for Custom sequence processing for 3d conv network. More...

struct	CustomTensorParams
	Tensor params passed to custom library for tensor preparation. More...

struct	CustomInitParams
	Custom Initialization parameter for custom library. More...

struct	NvDsPreProcessCustomBuf
	Custom Buffer passed to the custom lib for preparing tensor. More...

class	NvDsPreProcessAcquirer
	class for acquiring and releasing a buffer from tensor pool by custom lib. More...

struct	NvDsPreProcessUnit
	A preprocess unit for processing which can be Frame/ROI. More...

struct	NvDsPreProcessBatch
	Holds information about the batch of frames to be inferred. More...

struct	NvDsPreProcessTensorMeta
	tensor meta containing prepared tensor and related info inside preprocess user meta which is attached at batch level More...

struct	GstNvDsPreProcessBatchMeta
	preprocess meta as a user meta which is attached at batch level More...

Macros
#define	_PATH_MAX 4096
	Maximum file path length. More...

#define	NVDSPREPROCESS_USER_CONFIGS_PIXEL_NORMALIZATION_FACTOR "pixel-normalization-factor"
	pixel-normalization-factor config parameter More...

#define	NVDSPREPROCESS_USER_CONFIGS_MEAN_FILE "mean-file"
	mean-file config parameter More...

#define	NVDSPREPROCESS_USER_CONFIGS_OFFSETS "offsets"
	offsets config parameter More...

Typedefs
typedef struct CustomCtx	CustomCtx
	Context for custom library. More...

typedef void(*	NvDsPreProcessConvertFcn) (float outBuffer, unsigned char inBuffer, unsigned int width, unsigned int height, unsigned int pitch, float scaleFactor, float *meanDataBuffer, cudaStream_t stream)
	Function pointer type to which any of the NvDsPreProcessConvert functions can be assigned. More...

typedef struct CustomCtx	CustomCtx
	Context for custom library. More...

Enumerations
enum	NvDsPreProcessStatus { NVDSPREPROCESS_SUCCESS = 0, NVDSPREPROCESS_CONFIG_FAILED, NVDSPREPROCESS_CUSTOM_LIB_FAILED, NVDSPREPROCESS_CUSTOM_TRANSFORMATION_FAILED, NVDSPREPROCESS_CUSTOM_TENSOR_FAILED, NVDSPREPROCESS_INVALID_PARAMS, NVDSPREPROCESS_OUTPUT_PARSING_FAILED, NVDSPREPROCESS_CUDA_ERROR, NVDSPREPROCESS_TENSORRT_ERROR, NVDSPREPROCESS_RESOURCE_ERROR, NVDSPREPROCESS_TENSOR_NOT_READY, NVDSPREPROCESS_SUCCESS = 0, NVDSPREPROCESS_CONFIG_FAILED, NVDSPREPROCESS_CUSTOM_LIB_FAILED, NVDSPREPROCESS_CUSTOM_TRANSFORMATION_FAILED, NVDSPREPROCESS_CUSTOM_TENSOR_FAILED, NVDSPREPROCESS_INVALID_PARAMS, NVDSPREPROCESS_OUTPUT_PARSING_FAILED, NVDSPREPROCESS_CUDA_ERROR, NVDSPREPROCESS_TENSORRT_ERROR, NVDSPREPROCESS_RESOURCE_ERROR, NVDSPREPROCESS_TENSOR_NOT_READY }
	Enum for the status codes returned by NvDsPreProcessImpl. More...

enum	NvDsPreProcessNetworkInputOrder { NvDsPreProcessNetworkInputOrder_kNCHW = 0, NvDsPreProcessNetworkInputOrder_kNHWC, NvDsPreProcessNetworkInputOrder_CUSTOM, NvDsPreProcessNetworkInputOrder_kNCHW = 0, NvDsPreProcessNetworkInputOrder_kNHWC, NvDsPreProcessNetworkInputOrder_CUSTOM }
	Enum for the network input order according to which network shape will be provided to prepare raw tensor for inferencing. More...

enum	NvDsPreProcessFormat { NvDsPreProcessFormat_RGB, NvDsPreProcessFormat_BGR, NvDsPreProcessFormat_GRAY, NvDsPreProcessFormat_RGBA, NvDsPreProcessFormat_BGRx, NvDsPreProcessFormat_Tensor, NvDsPreProcessFormat_Unknown = 0xFFFFFFFF, NvDsPreProcessFormat_RGB, NvDsPreProcessFormat_BGR, NvDsPreProcessFormat_GRAY, NvDsPreProcessFormat_RGBA, NvDsPreProcessFormat_BGRx, NvDsPreProcessFormat_Tensor, NvDsPreProcessFormat_Unknown = 0xFFFFFFFF }
	Defines model color formats. More...

enum	NvDsPreProcessStatus { NVDSPREPROCESS_SUCCESS = 0, NVDSPREPROCESS_CONFIG_FAILED, NVDSPREPROCESS_CUSTOM_LIB_FAILED, NVDSPREPROCESS_CUSTOM_TRANSFORMATION_FAILED, NVDSPREPROCESS_CUSTOM_TENSOR_FAILED, NVDSPREPROCESS_INVALID_PARAMS, NVDSPREPROCESS_OUTPUT_PARSING_FAILED, NVDSPREPROCESS_CUDA_ERROR, NVDSPREPROCESS_TENSORRT_ERROR, NVDSPREPROCESS_RESOURCE_ERROR, NVDSPREPROCESS_TENSOR_NOT_READY, NVDSPREPROCESS_SUCCESS = 0, NVDSPREPROCESS_CONFIG_FAILED, NVDSPREPROCESS_CUSTOM_LIB_FAILED, NVDSPREPROCESS_CUSTOM_TRANSFORMATION_FAILED, NVDSPREPROCESS_CUSTOM_TENSOR_FAILED, NVDSPREPROCESS_INVALID_PARAMS, NVDSPREPROCESS_OUTPUT_PARSING_FAILED, NVDSPREPROCESS_CUDA_ERROR, NVDSPREPROCESS_TENSORRT_ERROR, NVDSPREPROCESS_RESOURCE_ERROR, NVDSPREPROCESS_TENSOR_NOT_READY }
	Enum for the status codes returned by NvDsPreProcessImpl. More...

enum	NvDsPreProcessNetworkInputOrder { NvDsPreProcessNetworkInputOrder_kNCHW = 0, NvDsPreProcessNetworkInputOrder_kNHWC, NvDsPreProcessNetworkInputOrder_CUSTOM, NvDsPreProcessNetworkInputOrder_kNCHW = 0, NvDsPreProcessNetworkInputOrder_kNHWC, NvDsPreProcessNetworkInputOrder_CUSTOM }
	Enum for the network input order according to which network shape will be provided to prepare raw tensor for inferencing. More...

enum	NvDsPreProcessFormat { NvDsPreProcessFormat_RGB, NvDsPreProcessFormat_BGR, NvDsPreProcessFormat_GRAY, NvDsPreProcessFormat_RGBA, NvDsPreProcessFormat_BGRx, NvDsPreProcessFormat_Tensor, NvDsPreProcessFormat_Unknown = 0xFFFFFFFF, NvDsPreProcessFormat_RGB, NvDsPreProcessFormat_BGR, NvDsPreProcessFormat_GRAY, NvDsPreProcessFormat_RGBA, NvDsPreProcessFormat_BGRx, NvDsPreProcessFormat_Tensor, NvDsPreProcessFormat_Unknown = 0xFFFFFFFF }
	Defines model color formats. More...

Functions
void	NvDsPreProcessConvert_C3ToP3Float (float outBuffer, unsigned char inBuffer, unsigned int width, unsigned int height, unsigned int pitch, float scaleFactor, float *meanDataBuffer, cudaStream_t stream)
	Converts an input packed 3 channel buffer of width x height resolution into an planar 3-channel float buffer of width x height resolution. More...

void	NvDsPreProcessConvert_C3ToL3Float (float outBuffer, unsigned char inBuffer, unsigned int width, unsigned int height, unsigned int pitch, float scaleFactor, float *meanDataBuffer, cudaStream_t stream)
	Converts an input packed 3 channel buffer of width x height resolution into an linear 3-channel float buffer of width x height resolution. More...

void	NvDsPreProcessConvert_C4ToP3Float (float outBuffer, unsigned char inBuffer, unsigned int width, unsigned int height, unsigned int pitch, float scaleFactor, float *meanDataBuffer, cudaStream_t stream)
	Converts an input packed 3 channel buffer of width x height resolution into an planar 3-channel float buffer of width x height resolution. More...

void	NvDsPreProcessConvert_C4ToL3Float (float outBuffer, unsigned char inBuffer, unsigned int width, unsigned int height, unsigned int pitch, float scaleFactor, float *meanDataBuffer, cudaStream_t stream)
	Converts an input packed 3 channel buffer of width x height resolution into an linear 3-channel float buffer of width x height resolution. More...

void	NvDsPreProcessConvert_C3ToP3RFloat (float outBuffer, unsigned char inBuffer, unsigned int width, unsigned int height, unsigned int pitch, float scaleFactor, float *meanDataBuffer, cudaStream_t stream)
	Converts an input packed 3 channel buffer of width x height resolution into an planar 3-channel float buffer of width x height resolution with plane order reversed. More...

void	NvDsPreProcessConvert_C3ToL3RFloat (float outBuffer, unsigned char inBuffer, unsigned int width, unsigned int height, unsigned int pitch, float scaleFactor, float *meanDataBuffer, cudaStream_t stream)
	Converts an input packed 3 channel buffer of width x height resolution into an linear 3-channel float buffer of width x height resolution with plane order reversed. More...

void	NvDsPreProcessConvert_C4ToP3RFloat (float outBuffer, unsigned char inBuffer, unsigned int width, unsigned int height, unsigned int pitch, float scaleFactor, float *meanDataBuffer, cudaStream_t stream)
	Converts an input packed 4 channel buffer of width x height resolution into an planar 3-channel float buffer of width x height resolution with plane order reversed. More...

void	NvDsPreProcessConvert_C4ToL3RFloat (float outBuffer, unsigned char inBuffer, unsigned int width, unsigned int height, unsigned int pitch, float scaleFactor, float *meanDataBuffer, cudaStream_t stream)
	Converts an input packed 4 channel buffer of width x height resolution into an linear 3-channel float buffer of width x height resolution with plane order reversed. More...

void	NvDsPreProcessConvert_C1ToP1Float (float outBuffer, unsigned char inBuffer, unsigned int width, unsigned int height, unsigned int pitch, float scaleFactor, float *meanDataBuffer, cudaStream_t stream)
	Converts an 1 channel UINT8 input of width x height resolution into an 1 channel float buffer of width x height resolution. More...

void	NvDsPreProcessConvert_FtFTensor (float outBuffer, float inBuffer, unsigned int width, unsigned int height, unsigned int pitch, float scaleFactor, float *meanDataBuffer, cudaStream_t stream)

NvDsPreProcessStatus	CustomTransformation (NvBufSurface in_surf, NvBufSurface out_surf, CustomTransformParams &params)
	Custom transformation function for group. More...

NvDsPreProcessStatus	CustomAsyncTransformation (NvBufSurface in_surf, NvBufSurface out_surf, CustomTransformParams &params)
	Custom Asynchronus group transformation function. More...

NvDsPreProcessStatus	CustomTensorPreparation (CustomCtx ctx, NvDsPreProcessBatch batch, NvDsPreProcessCustomBuf &buf, CustomTensorParams &tensorParam, NvDsPreProcessAcquirer acquirer)
	Custom tensor preparation function for NCHW/NHWC network order. More...

CustomCtx *	initLib (CustomInitParams initparams)
	custom library initialization function More...

void	deInitLib (CustomCtx *ctx)
	custom library deinitialization function More...

Variables
void *	NvDsPreProcessTensorMeta::raw_tensor_buffer
	raw tensor buffer preprocessed for infer More...

guint64	NvDsPreProcessTensorMeta::buffer_size
	size of raw tensor buffer More...

std::vector< int >	NvDsPreProcessTensorMeta::tensor_shape
	raw tensor buffer shape More...

NvDsDataType	NvDsPreProcessTensorMeta::data_type
	model datatype for which tensor prepared More...

std::string	NvDsPreProcessTensorMeta::tensor_name
	to be same as model input layer name More...

guint	NvDsPreProcessTensorMeta::gpu_id
	gpu-id on which tensor prepared More...

void *	NvDsPreProcessTensorMeta::private_data
	pointer to buffer from tensor pool More...

guint	NvDsPreProcessTensorMeta::meta_id
	meta id for differentiating between multiple tensor meta from same gst buffer,for the case when sum of roi's exceeds the batch size More...

std::vector< guint64 >	GstNvDsPreProcessBatchMeta::target_unique_ids
	target unique ids for which meta is prepared More...

NvDsPreProcessTensorMeta *	GstNvDsPreProcessBatchMeta::tensor_meta
	pointer to tensor meta More...

std::vector< NvDsRoiMeta >	GstNvDsPreProcessBatchMeta::roi_vector
	list of roi vectors per batch More...

void *	GstNvDsPreProcessBatchMeta::private_data
	pointer to buffer from scaling pool More...

Macro Definition Documentation

◆ _PATH_MAX

#define _PATH_MAX 4096

Maximum file path length.

Definition at line 46 of file nvdspreprocess_lib.h.

◆ NVDSPREPROCESS_USER_CONFIGS_MEAN_FILE

#define NVDSPREPROCESS_USER_CONFIGS_MEAN_FILE "mean-file"

mean-file config parameter

Definition at line 52 of file nvdspreprocess_lib.h.

◆ NVDSPREPROCESS_USER_CONFIGS_OFFSETS

#define NVDSPREPROCESS_USER_CONFIGS_OFFSETS "offsets"

offsets config parameter

Definition at line 55 of file nvdspreprocess_lib.h.

◆ NVDSPREPROCESS_USER_CONFIGS_PIXEL_NORMALIZATION_FACTOR

#define NVDSPREPROCESS_USER_CONFIGS_PIXEL_NORMALIZATION_FACTOR "pixel-normalization-factor"

pixel-normalization-factor config parameter

Definition at line 49 of file nvdspreprocess_lib.h.

Typedef Documentation

◆ CustomCtx [1/2]

typedef struct CustomCtx CustomCtx

Context for custom library.

Definition at line 54 of file gst-plugins/gst-nvdspreprocess/include/nvdspreprocess_interface.h.

◆ CustomCtx [2/2]

typedef struct CustomCtx CustomCtx

Context for custom library.

Definition at line 54 of file includes/nvdspreprocess_interface.h.

◆ NvDsPreProcessConvertFcn

typedef void(* NvDsPreProcessConvertFcn) (float *outBuffer, unsigned char *inBuffer, unsigned int width, unsigned int height, unsigned int pitch, float scaleFactor, float *meanDataBuffer, cudaStream_t stream)

Function pointer type to which any of the NvDsPreProcessConvert functions can be assigned.

Definition at line 334 of file nvdspreprocess_conversion.h.

Enumeration Type Documentation

◆ NvDsPreProcessFormat [1/2]

enum NvDsPreProcessFormat

Defines model color formats.

Enumerator
NvDsPreProcessFormat_RGB	Specifies 24-bit interleaved R-G-B format.
NvDsPreProcessFormat_BGR	Specifies 24-bit interleaved B-G-R format.
NvDsPreProcessFormat_GRAY	Specifies 8-bit Luma format.
NvDsPreProcessFormat_RGBA	Specifies 32-bit interleaved R-G-B-A format.
NvDsPreProcessFormat_BGRx	Specifies 32-bit interleaved B-G-R-x format.
NvDsPreProcessFormat_Tensor	NCHW planar.
NvDsPreProcessFormat_Unknown
NvDsPreProcessFormat_RGB	Specifies 24-bit interleaved R-G-B format.
NvDsPreProcessFormat_BGR	Specifies 24-bit interleaved B-G-R format.
NvDsPreProcessFormat_GRAY	Specifies 8-bit Luma format.
NvDsPreProcessFormat_RGBA	Specifies 32-bit interleaved R-G-B-A format.
NvDsPreProcessFormat_BGRx	Specifies 32-bit interleaved B-G-R-x format.
NvDsPreProcessFormat_Tensor	NCHW planar.
NvDsPreProcessFormat_Unknown

Definition at line 104 of file gst-plugins/gst-nvdspreprocess/include/nvdspreprocess_interface.h.

◆ NvDsPreProcessFormat [2/2]

enum NvDsPreProcessFormat

Defines model color formats.

Enumerator
NvDsPreProcessFormat_RGB	Specifies 24-bit interleaved R-G-B format.
NvDsPreProcessFormat_BGR	Specifies 24-bit interleaved B-G-R format.
NvDsPreProcessFormat_GRAY	Specifies 8-bit Luma format.
NvDsPreProcessFormat_RGBA	Specifies 32-bit interleaved R-G-B-A format.
NvDsPreProcessFormat_BGRx	Specifies 32-bit interleaved B-G-R-x format.
NvDsPreProcessFormat_Tensor	NCHW planar.
NvDsPreProcessFormat_Unknown
NvDsPreProcessFormat_RGB	Specifies 24-bit interleaved R-G-B format.
NvDsPreProcessFormat_BGR	Specifies 24-bit interleaved B-G-R format.
NvDsPreProcessFormat_GRAY	Specifies 8-bit Luma format.
NvDsPreProcessFormat_RGBA	Specifies 32-bit interleaved R-G-B-A format.
NvDsPreProcessFormat_BGRx	Specifies 32-bit interleaved B-G-R-x format.
NvDsPreProcessFormat_Tensor	NCHW planar.
NvDsPreProcessFormat_Unknown

Definition at line 104 of file includes/nvdspreprocess_interface.h.

◆ NvDsPreProcessNetworkInputOrder [1/2]

enum NvDsPreProcessNetworkInputOrder

Enum for the network input order according to which network shape will be provided to prepare raw tensor for inferencing.

Enumerator
NvDsPreProcessNetworkInputOrder_kNCHW	Specifies NCHW network input order.
NvDsPreProcessNetworkInputOrder_kNHWC	Specifies NHWC network input order.
NvDsPreProcessNetworkInputOrder_CUSTOM	Specifies any other custom input order handled by custom lib.
NvDsPreProcessNetworkInputOrder_kNCHW	Specifies NCHW network input order.
NvDsPreProcessNetworkInputOrder_kNHWC	Specifies NHWC network input order.
NvDsPreProcessNetworkInputOrder_CUSTOM	Specifies any other custom input order handled by custom lib.

Definition at line 91 of file includes/nvdspreprocess_interface.h.

◆ NvDsPreProcessNetworkInputOrder [2/2]

enum NvDsPreProcessNetworkInputOrder

Enum for the network input order according to which network shape will be provided to prepare raw tensor for inferencing.

Enumerator
NvDsPreProcessNetworkInputOrder_kNCHW	Specifies NCHW network input order.
NvDsPreProcessNetworkInputOrder_kNHWC	Specifies NHWC network input order.
NvDsPreProcessNetworkInputOrder_CUSTOM	Specifies any other custom input order handled by custom lib.
NvDsPreProcessNetworkInputOrder_kNCHW	Specifies NCHW network input order.
NvDsPreProcessNetworkInputOrder_kNHWC	Specifies NHWC network input order.
NvDsPreProcessNetworkInputOrder_CUSTOM	Specifies any other custom input order handled by custom lib.

Definition at line 91 of file gst-plugins/gst-nvdspreprocess/include/nvdspreprocess_interface.h.

◆ NvDsPreProcessStatus [1/2]

enum NvDsPreProcessStatus

Enum for the status codes returned by NvDsPreProcessImpl.

Enumerator
NVDSPREPROCESS_SUCCESS	NvDsPreprocess operation succeeded.
NVDSPREPROCESS_CONFIG_FAILED	Failed to configure the tensor_impl instance possibly due to an erroneous initialization property.
NVDSPREPROCESS_CUSTOM_LIB_FAILED	Custom Library interface implementation failed.
NVDSPREPROCESS_CUSTOM_TRANSFORMATION_FAILED	Custom Group Transformation failed.
NVDSPREPROCESS_CUSTOM_TENSOR_FAILED	Custom Tensor Preparation failed.
NVDSPREPROCESS_INVALID_PARAMS	Invalid parameters were supplied.
NVDSPREPROCESS_OUTPUT_PARSING_FAILED	Output parsing failed.
NVDSPREPROCESS_CUDA_ERROR	CUDA error was encountered.
NVDSPREPROCESS_TENSORRT_ERROR	TensorRT interface failed.
NVDSPREPROCESS_RESOURCE_ERROR	Resource error was encountered.
NVDSPREPROCESS_TENSOR_NOT_READY	Tensor Yet not ready to be attached as meta.
NVDSPREPROCESS_SUCCESS	NvDsPreprocess operation succeeded.
NVDSPREPROCESS_CONFIG_FAILED	Failed to configure the tensor_impl instance possibly due to an erroneous initialization property.
NVDSPREPROCESS_CUSTOM_LIB_FAILED	Custom Library interface implementation failed.
NVDSPREPROCESS_CUSTOM_TRANSFORMATION_FAILED	Custom Group Transformation failed.
NVDSPREPROCESS_CUSTOM_TENSOR_FAILED	Custom Tensor Preparation failed.
NVDSPREPROCESS_INVALID_PARAMS	Invalid parameters were supplied.
NVDSPREPROCESS_OUTPUT_PARSING_FAILED	Output parsing failed.
NVDSPREPROCESS_CUDA_ERROR	CUDA error was encountered.
NVDSPREPROCESS_TENSORRT_ERROR	TensorRT interface failed.
NVDSPREPROCESS_RESOURCE_ERROR	Resource error was encountered.
NVDSPREPROCESS_TENSOR_NOT_READY	Tensor Yet not ready to be attached as meta.

Definition at line 59 of file includes/nvdspreprocess_interface.h.

◆ NvDsPreProcessStatus [2/2]

enum NvDsPreProcessStatus

Enum for the status codes returned by NvDsPreProcessImpl.

Enumerator
NVDSPREPROCESS_SUCCESS	NvDsPreprocess operation succeeded.
NVDSPREPROCESS_CONFIG_FAILED	Failed to configure the tensor_impl instance possibly due to an erroneous initialization property.
NVDSPREPROCESS_CUSTOM_LIB_FAILED	Custom Library interface implementation failed.
NVDSPREPROCESS_CUSTOM_TRANSFORMATION_FAILED	Custom Group Transformation failed.
NVDSPREPROCESS_CUSTOM_TENSOR_FAILED	Custom Tensor Preparation failed.
NVDSPREPROCESS_INVALID_PARAMS	Invalid parameters were supplied.
NVDSPREPROCESS_OUTPUT_PARSING_FAILED	Output parsing failed.
NVDSPREPROCESS_CUDA_ERROR	CUDA error was encountered.
NVDSPREPROCESS_TENSORRT_ERROR	TensorRT interface failed.
NVDSPREPROCESS_RESOURCE_ERROR	Resource error was encountered.
NVDSPREPROCESS_TENSOR_NOT_READY	Tensor Yet not ready to be attached as meta.
NVDSPREPROCESS_SUCCESS	NvDsPreprocess operation succeeded.
NVDSPREPROCESS_CONFIG_FAILED	Failed to configure the tensor_impl instance possibly due to an erroneous initialization property.
NVDSPREPROCESS_CUSTOM_LIB_FAILED	Custom Library interface implementation failed.
NVDSPREPROCESS_CUSTOM_TRANSFORMATION_FAILED	Custom Group Transformation failed.
NVDSPREPROCESS_CUSTOM_TENSOR_FAILED	Custom Tensor Preparation failed.
NVDSPREPROCESS_INVALID_PARAMS	Invalid parameters were supplied.
NVDSPREPROCESS_OUTPUT_PARSING_FAILED	Output parsing failed.
NVDSPREPROCESS_CUDA_ERROR	CUDA error was encountered.
NVDSPREPROCESS_TENSORRT_ERROR	TensorRT interface failed.
NVDSPREPROCESS_RESOURCE_ERROR	Resource error was encountered.
NVDSPREPROCESS_TENSOR_NOT_READY	Tensor Yet not ready to be attached as meta.

Definition at line 59 of file gst-plugins/gst-nvdspreprocess/include/nvdspreprocess_interface.h.

Function Documentation

◆ CustomAsyncTransformation()

NvDsPreProcessStatus CustomAsyncTransformation	(	NvBufSurface *	in_surf,
		NvBufSurface *	out_surf,
		CustomTransformParams &	params
	)

Custom Asynchronus group transformation function.

◆ CustomTensorPreparation()

NvDsPreProcessStatus CustomTensorPreparation	(	CustomCtx *	ctx,
		NvDsPreProcessBatch *	batch,
		NvDsPreProcessCustomBuf *&	buf,
		CustomTensorParams &	tensorParam,
		NvDsPreProcessAcquirer *	acquirer
	)

Custom tensor preparation function for NCHW/NHWC network order.

◆ CustomTransformation()

NvDsPreProcessStatus CustomTransformation	(	NvBufSurface *	in_surf,
		NvBufSurface *	out_surf,
		CustomTransformParams &	params
	)

Custom transformation function for group.

◆ deInitLib()

void deInitLib ( CustomCtx * ctx )

custom library deinitialization function

◆ initLib()

CustomCtx* initLib ( CustomInitParams initparams )

custom library initialization function

◆ NvDsPreProcessConvert_C1ToP1Float()

void NvDsPreProcessConvert_C1ToP1Float	(	float *	outBuffer,
		unsigned char *	inBuffer,
		unsigned int	width,
		unsigned int	height,
		unsigned int	pitch,
		float	scaleFactor,
		float *	meanDataBuffer,
		cudaStream_t	stream
	)

Converts an 1 channel UINT8 input of width x height resolution into an 1 channel float buffer of width x height resolution.

The input buffer can have a pitch > width . The cuda kernel supports normalization and mean image subtraction.

Parameters

outBuffer	Cuda device buffer for float output. Should be at least (width * height * sizeof(float)) bytes.
inBuffer	Cuda device buffer for UINT8 input. Should be at least (pitch * height) bytes.
width	Width of the buffers in pixels.
height	Height of the buffers in pixels.
pitch	Pitch of the input buffer in bytes.
scaleFactor	Normalization factor.
meanDataBuffer	Mean Image Data buffer. Should be at least (width * height * sizeof(float)) bytes.
stream	Cuda stream identifier.

◆ NvDsPreProcessConvert_C3ToL3Float()

void NvDsPreProcessConvert_C3ToL3Float	(	float *	outBuffer,
		unsigned char *	inBuffer,
		unsigned int	width,
		unsigned int	height,
		unsigned int	pitch,
		float	scaleFactor,
		float *	meanDataBuffer,
		cudaStream_t	stream
	)

Converts an input packed 3 channel buffer of width x height resolution into an linear 3-channel float buffer of width x height resolution.

The input buffer can have a pitch > (width * 3). The cuda kernel supports normalization and mean image subtraction.

This kernel can be used for RGB -> RGB and BGR -> BGR conversions.

Parameters

outBuffer	Cuda device buffer for planar float output. Should be at least (width * height * 3 * sizeof(float)) bytes.
inBuffer	Cuda device buffer for packed input. Should be at least (pitch * height) bytes.
width	Width of the buffers in pixels.
height	Height of the buffers in pixels.
pitch	Pitch of the input buffer in bytes.
scaleFactor	Normalization factor.
meanDataBuffer	Mean Image Data buffer. Should be at least (width * height * 3 * sizeof(float)) bytes.
stream	Cuda stream identifier.

◆ NvDsPreProcessConvert_C3ToL3RFloat()

void NvDsPreProcessConvert_C3ToL3RFloat	(	float *	outBuffer,
		unsigned char *	inBuffer,
		unsigned int	width,
		unsigned int	height,
		unsigned int	pitch,
		float	scaleFactor,
		float *	meanDataBuffer,
		cudaStream_t	stream
	)

Converts an input packed 3 channel buffer of width x height resolution into an linear 3-channel float buffer of width x height resolution with plane order reversed.

The input buffer can have a pitch > (width * 3). The cuda kernel supports normalization and mean image subtraction.

This kernel can be used for BGR -> RGB and RGB -> BGR conversions.

Parameters

outBuffer	Cuda device buffer for planar float output. Should be at least (width * height * 3 * sizeof(float)) bytes.
inBuffer	Cuda device buffer for packed input. Should be at least (pitch * height) bytes.
width	Width of the buffers in pixels.
height	Height of the buffers in pixels.
pitch	Pitch of the input buffer in bytes.
scaleFactor	Normalization factor.
meanDataBuffer	Mean Image Data buffer. Should be at least (width * height * 3 * sizeof(float)) bytes.
stream	Cuda stream identifier.

◆ NvDsPreProcessConvert_C3ToP3Float()

void NvDsPreProcessConvert_C3ToP3Float	(	float *	outBuffer,
		unsigned char *	inBuffer,
		unsigned int	width,
		unsigned int	height,
		unsigned int	pitch,
		float	scaleFactor,
		float *	meanDataBuffer,
		cudaStream_t	stream
	)

Converts an input packed 3 channel buffer of width x height resolution into an planar 3-channel float buffer of width x height resolution.

The input buffer can have a pitch > (width * 3). The cuda kernel supports normalization and mean image subtraction.

This kernel can be used for RGB -> RGB and BGR -> BGR conversions.

Parameters

outBuffer	Cuda device buffer for planar float output. Should be at least (width * height * 3 * sizeof(float)) bytes.
inBuffer	Cuda device buffer for packed input. Should be at least (pitch * height) bytes.
width	Width of the buffers in pixels.
height	Height of the buffers in pixels.
pitch	Pitch of the input buffer in bytes.
scaleFactor	Normalization factor.
meanDataBuffer	Mean Image Data buffer. Should be at least (width * height * 3 * sizeof(float)) bytes.
stream	Cuda stream identifier.

◆ NvDsPreProcessConvert_C3ToP3RFloat()

void NvDsPreProcessConvert_C3ToP3RFloat	(	float *	outBuffer,
		unsigned char *	inBuffer,
		unsigned int	width,
		unsigned int	height,
		unsigned int	pitch,
		float	scaleFactor,
		float *	meanDataBuffer,
		cudaStream_t	stream
	)

Converts an input packed 3 channel buffer of width x height resolution into an planar 3-channel float buffer of width x height resolution with plane order reversed.

The input buffer can have a pitch > (width * 3). The cuda kernel supports normalization and mean image subtraction.

This kernel can be used for BGR -> RGB and RGB -> BGR conversions.

Parameters

outBuffer	Cuda device buffer for planar float output. Should be at least (width * height * 3 * sizeof(float)) bytes.
inBuffer	Cuda device buffer for packed input. Should be at least (pitch * height) bytes.
width	Width of the buffers in pixels.
height	Height of the buffers in pixels.
pitch	Pitch of the input buffer in bytes.
scaleFactor	Normalization factor.
meanDataBuffer	Mean Image Data buffer. Should be at least (width * height * 3 * sizeof(float)) bytes.
stream	Cuda stream identifier.

◆ NvDsPreProcessConvert_C4ToL3Float()

void NvDsPreProcessConvert_C4ToL3Float	(	float *	outBuffer,
		unsigned char *	inBuffer,
		unsigned int	width,
		unsigned int	height,
		unsigned int	pitch,
		float	scaleFactor,
		float *	meanDataBuffer,
		cudaStream_t	stream
	)

Converts an input packed 3 channel buffer of width x height resolution into an linear 3-channel float buffer of width x height resolution.

The input buffer can have a pitch > (width * 3). The cuda kernel supports normalization and mean image subtraction.

This kernel can be used for RGBA -> RGB and BGRx -> BGR conversions.

Parameters

outBuffer	Cuda device buffer for linear float output. Should be at least (width * height * 3 * sizeof(float)) bytes.
inBuffer	Cuda device buffer for packed input. Should be at least (pitch * height) bytes.
width	Width of the buffers in pixels.
height	Height of the buffers in pixels.
pitch	Pitch of the input buffer in bytes.
scaleFactor	Normalization factor.
meanDataBuffer	Mean Image Data buffer. Should be at least (width * height * 3 * sizeof(float)) bytes.
stream	Cuda stream identifier.

◆ NvDsPreProcessConvert_C4ToL3RFloat()

void NvDsPreProcessConvert_C4ToL3RFloat	(	float *	outBuffer,
		unsigned char *	inBuffer,
		unsigned int	width,
		unsigned int	height,
		unsigned int	pitch,
		float	scaleFactor,
		float *	meanDataBuffer,
		cudaStream_t	stream
	)

Converts an input packed 4 channel buffer of width x height resolution into an linear 3-channel float buffer of width x height resolution with plane order reversed.

The input buffer can have a pitch > (width * 3). The cuda kernel supports normalization and mean image subtraction.

This kernel can be used for BGRx -> RGB and RGBA -> BGR conversions.

Parameters

outBuffer	Cuda device buffer for planar float output. Should be at least (width * height * 3 * sizeof(float)) bytes.
inBuffer	Cuda device buffer for packed input. Should be at least (pitch * height) bytes.
width	Width of the buffers in pixels.
height	Height of the buffers in pixels.
pitch	Pitch of the input buffer in bytes.
scaleFactor	Normalization factor.
meanDataBuffer	Mean Image Data buffer. Should be at least (width * height * 3 * sizeof(float)) bytes.
stream	Cuda stream identifier.

◆ NvDsPreProcessConvert_C4ToP3Float()

void NvDsPreProcessConvert_C4ToP3Float	(	float *	outBuffer,
		unsigned char *	inBuffer,
		unsigned int	width,
		unsigned int	height,
		unsigned int	pitch,
		float	scaleFactor,
		float *	meanDataBuffer,
		cudaStream_t	stream
	)

Converts an input packed 3 channel buffer of width x height resolution into an planar 3-channel float buffer of width x height resolution.

The input buffer can have a pitch > (width * 3). The cuda kernel supports normalization and mean image subtraction.

This kernel can be used for RGBA -> RGB and BGRx -> BGR conversions.

Parameters

outBuffer	Cuda device buffer for planar float output. Should be at least (width * height * 3 * sizeof(float)) bytes.
inBuffer	Cuda device buffer for packed input. Should be at least (pitch * height) bytes.
width	Width of the buffers in pixels.
height	Height of the buffers in pixels.
pitch	Pitch of the input buffer in bytes.
scaleFactor	Normalization factor.
meanDataBuffer	Mean Image Data buffer. Should be at least (width * height * 3 * sizeof(float)) bytes.
stream	Cuda stream identifier.

◆ NvDsPreProcessConvert_C4ToP3RFloat()

void NvDsPreProcessConvert_C4ToP3RFloat	(	float *	outBuffer,
		unsigned char *	inBuffer,
		unsigned int	width,
		unsigned int	height,
		unsigned int	pitch,
		float	scaleFactor,
		float *	meanDataBuffer,
		cudaStream_t	stream
	)

Converts an input packed 4 channel buffer of width x height resolution into an planar 3-channel float buffer of width x height resolution with plane order reversed.

The input buffer can have a pitch > (width * 3). The cuda kernel supports normalization and mean image subtraction.

This kernel can be used for BGRx -> RGB and RGBA -> BGR conversions.

Parameters

outBuffer	Cuda device buffer for planar float output. Should be at least (width * height * 3 * sizeof(float)) bytes.
inBuffer	Cuda device buffer for packed input. Should be at least (pitch * height) bytes.
width	Width of the buffers in pixels.
height	Height of the buffers in pixels.
pitch	Pitch of the input buffer in bytes.
scaleFactor	Normalization factor.
meanDataBuffer	Mean Image Data buffer. Should be at least (width * height * 3 * sizeof(float)) bytes.
stream	Cuda stream identifier.

◆ NvDsPreProcessConvert_FtFTensor()

void NvDsPreProcessConvert_FtFTensor	(	float *	outBuffer,
		float *	inBuffer,
		unsigned int	width,
		unsigned int	height,
		unsigned int	pitch,
		float	scaleFactor,
		float *	meanDataBuffer,
		cudaStream_t	stream
	)

Variable Documentation

◆ buffer_size

guint64 NvDsPreProcessTensorMeta::buffer_size

size of raw tensor buffer

Definition at line 56 of file gst-plugins/gst-nvdspreprocess/include/nvdspreprocess_meta.h.

◆ data_type

NvDsDataType NvDsPreProcessTensorMeta::data_type

model datatype for which tensor prepared

Definition at line 62 of file gst-plugins/gst-nvdspreprocess/include/nvdspreprocess_meta.h.

◆ gpu_id

guint NvDsPreProcessTensorMeta::gpu_id

gpu-id on which tensor prepared

Definition at line 68 of file gst-plugins/gst-nvdspreprocess/include/nvdspreprocess_meta.h.

◆ meta_id

guint NvDsPreProcessTensorMeta::meta_id

meta id for differentiating between multiple tensor meta from same gst buffer,for the case when sum of roi's exceeds the batch size

Definition at line 74 of file gst-plugins/gst-nvdspreprocess/include/nvdspreprocess_meta.h.

◆ private_data [1/2]

void * NvDsPreProcessTensorMeta::private_data

pointer to buffer from tensor pool

Definition at line 71 of file gst-plugins/gst-nvdspreprocess/include/nvdspreprocess_meta.h.

◆ private_data [2/2]

void * GstNvDsPreProcessBatchMeta::private_data

pointer to buffer from scaling pool

Definition at line 94 of file gst-plugins/gst-nvdspreprocess/include/nvdspreprocess_meta.h.

◆ raw_tensor_buffer

void * NvDsPreProcessTensorMeta::raw_tensor_buffer

raw tensor buffer preprocessed for infer

Definition at line 53 of file gst-plugins/gst-nvdspreprocess/include/nvdspreprocess_meta.h.

◆ roi_vector

std::vector< NvDsRoiMeta > GstNvDsPreProcessBatchMeta::roi_vector

list of roi vectors per batch

Definition at line 91 of file gst-plugins/gst-nvdspreprocess/include/nvdspreprocess_meta.h.

◆ target_unique_ids

std::vector< guint64 > GstNvDsPreProcessBatchMeta::target_unique_ids

target unique ids for which meta is prepared

Definition at line 85 of file gst-plugins/gst-nvdspreprocess/include/nvdspreprocess_meta.h.

◆ tensor_meta

NvDsPreProcessTensorMeta * GstNvDsPreProcessBatchMeta::tensor_meta

pointer to tensor meta

Definition at line 88 of file gst-plugins/gst-nvdspreprocess/include/nvdspreprocess_meta.h.

◆ tensor_name

std::string NvDsPreProcessTensorMeta::tensor_name

to be same as model input layer name

Definition at line 65 of file gst-plugins/gst-nvdspreprocess/include/nvdspreprocess_meta.h.

◆ tensor_shape

std::vector< int > NvDsPreProcessTensorMeta::tensor_shape

raw tensor buffer shape

Definition at line 59 of file gst-plugins/gst-nvdspreprocess/include/nvdspreprocess_meta.h.


NVIDIA DeepStream SDK API Reference	6.3 Release

NVIDIA DeepStream SDK API Reference

6.3 Release

Detailed Description

Data Structures

Macros

Typedefs

Enumerations

Functions

Variables

Macro Definition Documentation

◆ _PATH_MAX

◆ NVDSPREPROCESS_USER_CONFIGS_MEAN_FILE

◆ NVDSPREPROCESS_USER_CONFIGS_OFFSETS

◆ NVDSPREPROCESS_USER_CONFIGS_PIXEL_NORMALIZATION_FACTOR

Typedef Documentation

◆ CustomCtx [1/2]

◆ CustomCtx [2/2]

◆ NvDsPreProcessConvertFcn

Enumeration Type Documentation

◆ NvDsPreProcessFormat [1/2]

◆ NvDsPreProcessFormat [2/2]

◆ NvDsPreProcessNetworkInputOrder [1/2]

◆ NvDsPreProcessNetworkInputOrder [2/2]

◆ NvDsPreProcessStatus [1/2]

◆ NvDsPreProcessStatus [2/2]

Function Documentation

◆ CustomAsyncTransformation()

◆ CustomTensorPreparation()

◆ CustomTransformation()

◆ deInitLib()

◆ initLib()

◆ NvDsPreProcessConvert_C1ToP1Float()

◆ NvDsPreProcessConvert_C3ToL3Float()

◆ NvDsPreProcessConvert_C3ToL3RFloat()

◆ NvDsPreProcessConvert_C3ToP3Float()

◆ NvDsPreProcessConvert_C3ToP3RFloat()

◆ NvDsPreProcessConvert_C4ToL3Float()

◆ NvDsPreProcessConvert_C4ToL3RFloat()

◆ NvDsPreProcessConvert_C4ToP3Float()

◆ NvDsPreProcessConvert_C4ToP3RFloat()

◆ NvDsPreProcessConvert_FtFTensor()

Variable Documentation

◆ buffer_size

◆ data_type

◆ gpu_id

◆ meta_id

◆ private_data [1/2]

◆ private_data [2/2]

◆ raw_tensor_buffer

◆ roi_vector

◆ target_unique_ids

◆ tensor_meta

◆ tensor_name

◆ tensor_shape