class InferenceEngine::IPreProcessData¶

Overview¶

This class stores pre-process information for exact input. More…

#include <ie_preprocess_data.hpp>

class IPreProcessData: public std::enable_shared_from_this< IPreProcessData >
{
public:
    // methods

    virtual void setRoiBlob(const Blob::Ptr& blob) = 0;
    virtual Blob::Ptr getRoiBlob() const = 0;

    virtual void execute(
        Blob::Ptr& preprocessedBlob,
        const PreProcessInfo& info,
        bool serial,
        int batchSize = -1
        ) = 0;

    virtual void isApplicable(const Blob::Ptr& src, const Blob::Ptr& dst) = 0;

protected:
};

// direct descendants

class PreProcessData;

Detailed Documentation¶

This class stores pre-process information for exact input.

Methods¶

virtual void setRoiBlob(const Blob::Ptr& blob) = 0

Sets ROI blob to be resized and placed to the default input blob during pre-processing.

Parameters:

blob

ROI blob.

virtual Blob::Ptr getRoiBlob() const = 0

Gets pointer to the ROI blob used for a given input.

Returns:

Blob pointer.

virtual void execute(
    Blob::Ptr& preprocessedBlob,
    const PreProcessInfo& info,
    bool serial,
    int batchSize = -1
    ) = 0

Executes input pre-processing with a given pre-processing information.

Parameters:

outBlob	pre-processed output blob to be used for inference.
info	pre-processing info that specifies resize algorithm and color format.
serial	disable OpenMP threading if the value set to true.
batchSize	batch size for pre-processing.