NvDsPreProcessTensorImpl Class Reference

Detailed Description

Provides pre-processing functionality like mean subtraction and normalization.

Definition at line 183 of file nvdspreprocess_impl.h.

Public Member Functions
	NvDsPreProcessTensorImpl (const NvDsPreProcessNetworkSize &size, NvDsPreProcessFormat format, int id=0)
	constructor for tensor preparation implementation More...
virtual	~NvDsPreProcessTensorImpl ()=default
bool	setScaleOffsets (float scale, const std::vector< float > &offsets={})
	method to set offsets values More...
bool	setMeanFile (const std::string &file)
	method to set mean file More...
bool	setInputOrder (const NvDsPreProcessNetworkInputOrder order)
	method to set network input order More...
NvDsPreProcessStatus	allocateResource ()
	allocate resources for tensor preparation More...
NvDsPreProcessStatus	syncStream ()
	synchronize cuda stream More...
NvDsPreProcessStatus	prepare_tensor (NvDsPreProcessBatch *batch, void *&devBuf)
	method to prepare tensor using cuda kernels More...

Constructor & Destructor Documentation

NvDsPreProcessTensorImpl()

NvDsPreProcessTensorImpl::NvDsPreProcessTensorImpl	(	const NvDsPreProcessNetworkSize &	size,
		NvDsPreProcessFormat	format,
		int	id = 0
	)

constructor for tensor preparation implementation

~NvDsPreProcessTensorImpl()

virtual NvDsPreProcessTensorImpl::~NvDsPreProcessTensorImpl ( )

virtualdefault

Member Function Documentation

allocateResource()

NvDsPreProcessStatus NvDsPreProcessTensorImpl::allocateResource

(

)

allocate resources for tensor preparation

prepare_tensor()

NvDsPreProcessStatus NvDsPreProcessTensorImpl::prepare_tensor	(	NvDsPreProcessBatch *	batch,
		void *&	devBuf
	)

method to prepare tensor using cuda kernels

setInputOrder()

bool NvDsPreProcessTensorImpl::setInputOrder

(

const NvDsPreProcessNetworkInputOrder

order

)

method to set network input order

setMeanFile()

bool NvDsPreProcessTensorImpl::setMeanFile

(

const std::string &

file

)

method to set mean file

setScaleOffsets()

bool NvDsPreProcessTensorImpl::setScaleOffsets	(	float	scale,
		const std::vector< float > &	offsets = {}
	)

method to set offsets values

syncStream()

NvDsPreProcessStatus NvDsPreProcessTensorImpl::syncStream

(

)

synchronize cuda stream

---

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

• nvdspreprocess_impl.h