Preprocessor object for PaddleClas serials model. More...

#include <rec_preprocessor.h>

Inheritance diagram for fastdeploy::vision::ocr::RecognizerPreprocessor:

Collaboration diagram for fastdeploy::vision::ocr::RecognizerPreprocessor:

Public Member Functions
bool	Run (std::vector< FDMat > images, std::vector< FDTensor > outputs, size_t start_index, size_t end_index, const std::vector< int > &indices)
	Process the input image and prepare input tensors for runtime. More...

virtual bool	Apply (FDMatBatch image_batch, std::vector< FDTensor > outputs)
	Implement the virtual function of ProcessorManager, Apply() is the body of Run(). Apply() contains the main logic of preprocessing, Run() is called by users to execute preprocessing. More...

void	SetStaticShapeInfer (bool static_shape_infer)

bool	GetStaticShapeInfer () const
	Get static_shape_infer of the recognition preprocess.

void	SetNormalize (const std::vector< float > &mean, const std::vector< float > &std, bool is_scale)

void	SetRecImageShape (const std::vector< int > &rec_image_shape)
	Set rec_image_shape for the recognition preprocess.

std::vector< int >	GetRecImageShape ()
	Get rec_image_shape for the recognition preprocess.

void	DisableNormalize ()
	This function will disable normalize in preprocessing step.

void	DisablePermute ()
	This function will disable hwc2chw in preprocessing step.

Public Member Functions inherited from fastdeploy::vision::ProcessorManager
void	UseCuda (bool enable_cv_cuda=false, int gpu_id=-1)
	Use CUDA to boost the performance of processors. More...

bool	Run (std::vector< FDMat > images, std::vector< FDTensor > outputs)
	Process the input images and prepare input tensors for runtime. More...

Detailed Description

Preprocessor object for PaddleClas serials model.

Member Function Documentation

◆ Apply()

bool fastdeploy::vision::ocr::RecognizerPreprocessor::Apply	(	FDMatBatch *	image_batch,
		std::vector< FDTensor > *	outputs
	)

virtual

Implement the virtual function of ProcessorManager, Apply() is the body of Run(). Apply() contains the main logic of preprocessing, Run() is called by users to execute preprocessing.

Parameters

[in]	image_batch	The input image batch
[in]	outputs	The output tensors which will feed in runtime

Returns: true if the preprocess successed, otherwise false

Implements fastdeploy::vision::ProcessorManager.

◆ Run()

bool fastdeploy::vision::ocr::RecognizerPreprocessor::Run	(	std::vector< FDMat > *	images,
		std::vector< FDTensor > *	outputs,
		size_t	start_index,
		size_t	end_index,
		const std::vector< int > &	indices
	)

Process the input image and prepare input tensors for runtime.

Parameters

[in]	images	The input data list, all the elements are FDMat
[in]	outputs	The output tensors which will be fed into runtime

Returns: true if the preprocess successed, otherwise false

◆ SetNormalize()

void fastdeploy::vision::ocr::RecognizerPreprocessor::SetNormalize	(	const std::vector< float > &	mean,
		const std::vector< float > &	std,
		bool	is_scale
	)

inline

Set preprocess normalize parameters, please call this API to customize the normalize parameters, otherwise it will use the default normalize parameters.

◆ SetStaticShapeInfer()

void fastdeploy::vision::ocr::RecognizerPreprocessor::SetStaticShapeInfer ( bool static_shape_infer )