OpenCV  4.1.0
Open Source Computer Vision
Public Types | Public Member Functions | Static Public Member Functions | Public Attributes | List of all members
cv::HOGDescriptor Struct Reference

Implementation of HOG (Histogram of Oriented Gradients) descriptor and object detector. More...

#include <opencv2/objdetect.hpp>

Public Types

enum  { DEFAULT_NLEVELS = 64 }
 
enum  DescriptorStorageFormat {
  DESCR_FORMAT_COL_BY_COL,
  DESCR_FORMAT_ROW_BY_ROW
}
 
enum  HistogramNormType { L2Hys = 0 }
 

Public Member Functions

 HOGDescriptor ()
 Creates the HOG descriptor and detector with default params.
 
 HOGDescriptor (Size _winSize, Size _blockSize, Size _blockStride, Size _cellSize, int _nbins, int _derivAperture=1, double _winSigma=-1, HOGDescriptor::HistogramNormType _histogramNormType=HOGDescriptor::L2Hys, double _L2HysThreshold=0.2, bool _gammaCorrection=false, int _nlevels=HOGDescriptor::DEFAULT_NLEVELS, bool _signedGradient=false)
 
 HOGDescriptor (const String &filename)
 
 HOGDescriptor (const HOGDescriptor &d)
 
virtual ~HOGDescriptor ()
 Default destructor.
 
bool checkDetectorSize () const
 Checks if detector size equal to descriptor size.
 
virtual void compute (InputArray img, std::vector< float > &descriptors, Size winStride=Size(), Size padding=Size(), const std::vector< Point > &locations=std::vector< Point >()) const
 Computes HOG descriptors of given image.
 
virtual void computeGradient (InputArray img, InputOutputArray grad, InputOutputArray angleOfs, Size paddingTL=Size(), Size paddingBR=Size()) const
 Computes gradients and quantized gradient orientations.
 
virtual void copyTo (HOGDescriptor &c) const
 clones the HOGDescriptor
 
virtual void detect (InputArray img, std::vector< Point > &foundLocations, std::vector< double > &weights, double hitThreshold=0, Size winStride=Size(), Size padding=Size(), const std::vector< Point > &searchLocations=std::vector< Point >()) const
 Performs object detection without a multi-scale window.
 
virtual void detect (InputArray img, std::vector< Point > &foundLocations, double hitThreshold=0, Size winStride=Size(), Size padding=Size(), const std::vector< Point > &searchLocations=std::vector< Point >()) const
 Performs object detection without a multi-scale window.
 
virtual void detectMultiScale (InputArray img, std::vector< Rect > &foundLocations, std::vector< double > &foundWeights, double hitThreshold=0, Size winStride=Size(), Size padding=Size(), double scale=1.05, double finalThreshold=2.0, bool useMeanshiftGrouping=false) const
 Detects objects of different sizes in the input image. The detected objects are returned as a list of rectangles.
 
virtual void detectMultiScale (InputArray img, std::vector< Rect > &foundLocations, double hitThreshold=0, Size winStride=Size(), Size padding=Size(), double scale=1.05, double finalThreshold=2.0, bool useMeanshiftGrouping=false) const
 Detects objects of different sizes in the input image. The detected objects are returned as a list of rectangles.
 
virtual void detectMultiScaleROI (InputArray img, std::vector< cv::Rect > &foundLocations, std::vector< DetectionROI > &locations, double hitThreshold=0, int groupThreshold=0) const
 evaluate specified ROI and return confidence value for each location in multiple scales
 
virtual void detectROI (InputArray img, const std::vector< cv::Point > &locations, std::vector< cv::Point > &foundLocations, std::vector< double > &confidences, double hitThreshold=0, cv::Size winStride=Size(), cv::Size padding=Size()) const
 evaluate specified ROI and return confidence value for each location
 
size_t getDescriptorSize () const
 Returns the number of coefficients required for the classification.
 
double getWinSigma () const
 Returns winSigma value.
 
void groupRectangles (std::vector< cv::Rect > &rectList, std::vector< double > &weights, int groupThreshold, double eps) const
 Groups the object candidate rectangles.
 
virtual bool load (const String &filename, const String &objname=String())
 loads HOGDescriptor parameters and coefficients for the linear SVM classifier from a file.
 
virtual bool read (FileNode &fn)
 Reads HOGDescriptor parameters from a cv::FileNode.
 
virtual void save (const String &filename, const String &objname=String()) const
 saves HOGDescriptor parameters and coefficients for the linear SVM classifier to a file
 
virtual void setSVMDetector (InputArray svmdetector)
 Sets coefficients for the linear SVM classifier.
 
virtual void write (FileStorage &fs, const String &objname) const
 Stores HOGDescriptor parameters in a cv::FileStorage.
 

Static Public Member Functions

static std::vector< float > getDaimlerPeopleDetector ()
 Returns coefficients of the classifier trained for people detection (for 48x96 windows).
 
static std::vector< float > getDefaultPeopleDetector ()
 Returns coefficients of the classifier trained for people detection (for 64x128 windows).
 

Public Attributes

Size blockSize
 Block size in pixels. Align to cell size. Default value is Size(16,16).
 
Size blockStride
 Block stride. It must be a multiple of cell size. Default value is Size(8,8).
 
Size cellSize
 Cell size. Default value is Size(8,8).
 
int derivAperture
 not documented
 
float free_coef
 not documented
 
bool gammaCorrection
 Flag to specify whether the gamma correction preprocessing is required or not.
 
HOGDescriptor::HistogramNormType histogramNormType
 histogramNormType
 
double L2HysThreshold
 L2-Hys normalization method shrinkage.
 
int nbins
 Number of bins used in the calculation of histogram of gradients. Default value is 9.
 
int nlevels
 Maximum number of detection window increases. Default value is 64.
 
UMat oclSvmDetector
 coefficients for the linear SVM classifier used when OpenCL is enabled
 
bool signedGradient
 Indicates signed gradient will be used or not.
 
std::vector< float > svmDetector
 coefficients for the linear SVM classifier.
 
double winSigma
 Gaussian smoothing window parameter.
 
Size winSize
 Detection window size. Align to block size and block stride. Default value is Size(64,128).
 

Detailed Description

Implementation of HOG (Histogram of Oriented Gradients) descriptor and object detector.

the HOG descriptor algorithm introduced by Navneet Dalal and Bill TriggsDalal2005 .

useful links:

https://hal.inria.fr/inria-00548512/document/

https://en.wikipedia.org/wiki/Histogram_of_oriented_gradients

https://software.intel.com/en-us/ipp-dev-reference-histogram-of-oriented-gradients-hog-descriptor

http://www.learnopencv.com/histogram-of-oriented-gradients

http://www.learnopencv.com/handwritten-digits-classification-an-opencv-c-python-tutorial

Examples:
samples/cpp/peopledetect.cpp, samples/cpp/train_HOG.cpp, and samples/tapi/hog.cpp.

Member Enumeration Documentation

anonymous enum
Enumerator
DEFAULT_NLEVELS 

Default nlevels value.

Enumerator
DESCR_FORMAT_COL_BY_COL 
DESCR_FORMAT_ROW_BY_ROW 
Enumerator
L2Hys 

Default histogramNormType.

Constructor & Destructor Documentation

cv::HOGDescriptor::HOGDescriptor ( )
inline

Creates the HOG descriptor and detector with default params.

aqual to HOGDescriptor(Size(64,128), Size(16,16), Size(8,8), Size(8,8), 9 )

cv::HOGDescriptor::HOGDescriptor ( Size  _winSize,
Size  _blockSize,
Size  _blockStride,
Size  _cellSize,
int  _nbins,
int  _derivAperture = 1,
double  _winSigma = -1,
HOGDescriptor::HistogramNormType  _histogramNormType = HOGDescriptor::L2Hys,
double  _L2HysThreshold = 0.2,
bool  _gammaCorrection = false,
int  _nlevels = HOGDescriptor::DEFAULT_NLEVELS,
bool  _signedGradient = false 
)
inline

This is an overloaded member function, provided for convenience. It differs from the above function only in what argument(s) it accepts.

Parameters
_winSizesets winSize with given value.
_blockSizesets blockSize with given value.
_blockStridesets blockStride with given value.
_cellSizesets cellSize with given value.
_nbinssets nbins with given value.
_derivAperturesets derivAperture with given value.
_winSigmasets winSigma with given value.
_histogramNormTypesets histogramNormType with given value.
_L2HysThresholdsets L2HysThreshold with given value.
_gammaCorrectionsets gammaCorrection with given value.
_nlevelssets nlevels with given value.
_signedGradientsets signedGradient with given value.
cv::HOGDescriptor::HOGDescriptor ( const String filename)
inline

This is an overloaded member function, provided for convenience. It differs from the above function only in what argument(s) it accepts.

Parameters
filenameThe file name containing HOGDescriptor properties and coefficients for the linear SVM classifier.
cv::HOGDescriptor::HOGDescriptor ( const HOGDescriptor d)
inline

This is an overloaded member function, provided for convenience. It differs from the above function only in what argument(s) it accepts.

Parameters
dthe HOGDescriptor which cloned to create a new one.
virtual cv::HOGDescriptor::~HOGDescriptor ( )
inlinevirtual

Default destructor.

Member Function Documentation

bool cv::HOGDescriptor::checkDetectorSize ( ) const

Checks if detector size equal to descriptor size.

virtual void cv::HOGDescriptor::compute ( InputArray  img,
std::vector< float > &  descriptors,
Size  winStride = Size(),
Size  padding = Size(),
const std::vector< Point > &  locations = std::vector< Point >() 
) const
virtual

Computes HOG descriptors of given image.

Parameters
imgMatrix of the type CV_8U containing an image where HOG features will be calculated.
descriptorsMatrix of the type CV_32F
winStrideWindow stride. It must be a multiple of block stride.
paddingPadding
locationsVector of Point
Examples:
samples/cpp/train_HOG.cpp.
virtual void cv::HOGDescriptor::computeGradient ( InputArray  img,
InputOutputArray  grad,
InputOutputArray  angleOfs,
Size  paddingTL = Size(),
Size  paddingBR = Size() 
) const
virtual

Computes gradients and quantized gradient orientations.

Parameters
imgMatrix contains the image to be computed
gradMatrix of type CV_32FC2 contains computed gradients
angleOfsMatrix of type CV_8UC2 contains quantized gradient orientations
paddingTLPadding from top-left
paddingBRPadding from bottom-right
virtual void cv::HOGDescriptor::copyTo ( HOGDescriptor c) const
virtual

clones the HOGDescriptor

Parameters
ccloned HOGDescriptor
virtual void cv::HOGDescriptor::detect ( InputArray  img,
std::vector< Point > &  foundLocations,
std::vector< double > &  weights,
double  hitThreshold = 0,
Size  winStride = Size(),
Size  padding = Size(),
const std::vector< Point > &  searchLocations = std::vector< Point >() 
) const
virtual

Performs object detection without a multi-scale window.

Parameters
imgMatrix of the type CV_8U or CV_8UC3 containing an image where objects are detected.
foundLocationsVector of point where each point contains left-top corner point of detected object boundaries.
weightsVector that will contain confidence values for each detected object.
hitThresholdThreshold for the distance between features and SVM classifying plane. Usually it is 0 and should be specified in the detector coefficients (as the last free coefficient). But if the free coefficient is omitted (which is allowed), you can specify it manually here.
winStrideWindow stride. It must be a multiple of block stride.
paddingPadding
searchLocationsVector of Point includes set of requested locations to be evaluated.
virtual void cv::HOGDescriptor::detect ( InputArray  img,
std::vector< Point > &  foundLocations,
double  hitThreshold = 0,
Size  winStride = Size(),
Size  padding = Size(),
const std::vector< Point > &  searchLocations = std::vector< Point >() 
) const
virtual

Performs object detection without a multi-scale window.

Parameters
imgMatrix of the type CV_8U or CV_8UC3 containing an image where objects are detected.
foundLocationsVector of point where each point contains left-top corner point of detected object boundaries.
hitThresholdThreshold for the distance between features and SVM classifying plane. Usually it is 0 and should be specified in the detector coefficients (as the last free coefficient). But if the free coefficient is omitted (which is allowed), you can specify it manually here.
winStrideWindow stride. It must be a multiple of block stride.
paddingPadding
searchLocationsVector of Point includes locations to search.
virtual void cv::HOGDescriptor::detectMultiScale ( InputArray  img,
std::vector< Rect > &  foundLocations,
std::vector< double > &  foundWeights,
double  hitThreshold = 0,
Size  winStride = Size(),
Size  padding = Size(),
double  scale = 1.05,
double  finalThreshold = 2.0,
bool  useMeanshiftGrouping = false 
) const
virtual

Detects objects of different sizes in the input image. The detected objects are returned as a list of rectangles.

Parameters
imgMatrix of the type CV_8U or CV_8UC3 containing an image where objects are detected.
foundLocationsVector of rectangles where each rectangle contains the detected object.
foundWeightsVector that will contain confidence values for each detected object.
hitThresholdThreshold for the distance between features and SVM classifying plane. Usually it is 0 and should be specified in the detector coefficients (as the last free coefficient). But if the free coefficient is omitted (which is allowed), you can specify it manually here.
winStrideWindow stride. It must be a multiple of block stride.
paddingPadding
scaleCoefficient of the detection window increase.
finalThresholdFinal threshold
useMeanshiftGroupingindicates grouping algorithm
Examples:
samples/cpp/train_HOG.cpp.
virtual void cv::HOGDescriptor::detectMultiScale ( InputArray  img,
std::vector< Rect > &  foundLocations,
double  hitThreshold = 0,
Size  winStride = Size(),
Size  padding = Size(),
double  scale = 1.05,
double  finalThreshold = 2.0,
bool  useMeanshiftGrouping = false 
) const
virtual

Detects objects of different sizes in the input image. The detected objects are returned as a list of rectangles.

Parameters
imgMatrix of the type CV_8U or CV_8UC3 containing an image where objects are detected.
foundLocationsVector of rectangles where each rectangle contains the detected object.
hitThresholdThreshold for the distance between features and SVM classifying plane. Usually it is 0 and should be specified in the detector coefficients (as the last free coefficient). But if the free coefficient is omitted (which is allowed), you can specify it manually here.
winStrideWindow stride. It must be a multiple of block stride.
paddingPadding
scaleCoefficient of the detection window increase.
finalThresholdFinal threshold
useMeanshiftGroupingindicates grouping algorithm
virtual void cv::HOGDescriptor::detectMultiScaleROI ( InputArray  img,
std::vector< cv::Rect > &  foundLocations,
std::vector< DetectionROI > &  locations,
double  hitThreshold = 0,
int  groupThreshold = 0 
) const
virtual

evaluate specified ROI and return confidence value for each location in multiple scales

Parameters
imgMatrix of the type CV_8U or CV_8UC3 containing an image where objects are detected.
foundLocationsVector of rectangles where each rectangle contains the detected object.
locationsVector of DetectionROI
hitThresholdThreshold for the distance between features and SVM classifying plane. Usually it is 0 and should be specified in the detector coefficients (as the last free coefficient). But if the free coefficient is omitted (which is allowed), you can specify it manually here.
groupThresholdMinimum possible number of rectangles minus 1. The threshold is used in a group of rectangles to retain it.
virtual void cv::HOGDescriptor::detectROI ( InputArray  img,
const std::vector< cv::Point > &  locations,
std::vector< cv::Point > &  foundLocations,
std::vector< double > &  confidences,
double  hitThreshold = 0,
cv::Size  winStride = Size(),
cv::Size  padding = Size() 
) const
virtual

evaluate specified ROI and return confidence value for each location

Parameters
imgMatrix of the type CV_8U or CV_8UC3 containing an image where objects are detected.
locationsVector of Point
foundLocationsVector of Point where each Point is detected object's top-left point.
confidencesconfidences
hitThresholdThreshold for the distance between features and SVM classifying plane. Usually it is 0 and should be specified in the detector coefficients (as the last free coefficient). But if the free coefficient is omitted (which is allowed), you can specify it manually here
winStridewinStride
paddingpadding
static std::vector<float> cv::HOGDescriptor::getDaimlerPeopleDetector ( )
static

Returns coefficients of the classifier trained for people detection (for 48x96 windows).

Examples:
samples/cpp/peopledetect.cpp.
static std::vector<float> cv::HOGDescriptor::getDefaultPeopleDetector ( )
static

Returns coefficients of the classifier trained for people detection (for 64x128 windows).

Examples:
samples/cpp/peopledetect.cpp.
size_t cv::HOGDescriptor::getDescriptorSize ( ) const

Returns the number of coefficients required for the classification.

double cv::HOGDescriptor::getWinSigma ( ) const

Returns winSigma value.

void cv::HOGDescriptor::groupRectangles ( std::vector< cv::Rect > &  rectList,
std::vector< double > &  weights,
int  groupThreshold,
double  eps 
) const

Groups the object candidate rectangles.

Parameters
rectListInput/output vector of rectangles. Output vector includes retained and grouped rectangles. (The Python list is not modified in place.)
weightsInput/output vector of weights of rectangles. Output vector includes weights of retained and grouped rectangles. (The Python list is not modified in place.)
groupThresholdMinimum possible number of rectangles minus 1. The threshold is used in a group of rectangles to retain it.
epsRelative difference between sides of the rectangles to merge them into a group.
virtual bool cv::HOGDescriptor::load ( const String filename,
const String objname = String() 
)
virtual

loads HOGDescriptor parameters and coefficients for the linear SVM classifier from a file.

Parameters
filenamePath of the file to read.
objnameThe optional name of the node to read (if empty, the first top-level node will be used).
Examples:
samples/cpp/train_HOG.cpp.
virtual bool cv::HOGDescriptor::read ( FileNode fn)
virtual

Reads HOGDescriptor parameters from a cv::FileNode.

Parameters
fnFile node
virtual void cv::HOGDescriptor::save ( const String filename,
const String objname = String() 
) const
virtual

saves HOGDescriptor parameters and coefficients for the linear SVM classifier to a file

Parameters
filenameFile name
objnameObject name
Examples:
samples/cpp/train_HOG.cpp.
virtual void cv::HOGDescriptor::setSVMDetector ( InputArray  svmdetector)
virtual

Sets coefficients for the linear SVM classifier.

Parameters
svmdetectorcoefficients for the linear SVM classifier.
Examples:
samples/cpp/train_HOG.cpp.
virtual void cv::HOGDescriptor::write ( FileStorage fs,
const String objname 
) const
virtual

Stores HOGDescriptor parameters in a cv::FileStorage.

Parameters
fsFile storage
objnameObject name

Member Data Documentation

Size cv::HOGDescriptor::blockSize

Block size in pixels. Align to cell size. Default value is Size(16,16).

Size cv::HOGDescriptor::blockStride

Block stride. It must be a multiple of cell size. Default value is Size(8,8).

Size cv::HOGDescriptor::cellSize

Cell size. Default value is Size(8,8).

int cv::HOGDescriptor::derivAperture

not documented

float cv::HOGDescriptor::free_coef

not documented

bool cv::HOGDescriptor::gammaCorrection

Flag to specify whether the gamma correction preprocessing is required or not.

HOGDescriptor::HistogramNormType cv::HOGDescriptor::histogramNormType

histogramNormType

double cv::HOGDescriptor::L2HysThreshold

L2-Hys normalization method shrinkage.

int cv::HOGDescriptor::nbins

Number of bins used in the calculation of histogram of gradients. Default value is 9.

int cv::HOGDescriptor::nlevels

Maximum number of detection window increases. Default value is 64.

UMat cv::HOGDescriptor::oclSvmDetector

coefficients for the linear SVM classifier used when OpenCL is enabled

bool cv::HOGDescriptor::signedGradient

Indicates signed gradient will be used or not.

std::vector<float> cv::HOGDescriptor::svmDetector

coefficients for the linear SVM classifier.

double cv::HOGDescriptor::winSigma

Gaussian smoothing window parameter.

Size cv::HOGDescriptor::winSize

Detection window size. Align to block size and block stride. Default value is Size(64,128).

Examples:
samples/cpp/train_HOG.cpp.

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