OpenCV  4.5.2 Open Source Computer Vision
Tracking API implementation details

## Namespaces

cv::detail::tracking::contrib_feature

cv::detail::tracking::kalman_filters

cv::detail::tracking::online_boosting

cv::detail::tracking::tbm

cv::detail::tracking::tld

## Classes

class  cv::detail::tracking::TrackerContribFeature
Abstract base class for TrackerContribFeature that represents the feature. More...

class  cv::detail::tracking::TrackerContribFeatureHAAR
TrackerContribFeature based on HAAR features, used by TrackerMIL and many others algorithms. More...

class  cv::detail::tracking::TrackerContribFeatureSet
Class that manages the extraction and selection of features. More...

class  cv::detail::tracking::TrackerContribSampler
Class that manages the sampler in order to select regions for the update the model of the tracker [AAM] Sampling e Labeling. See table I and section III B. More...

class  cv::detail::tracking::TrackerContribSamplerAlgorithm
Abstract base class for TrackerContribSamplerAlgorithm that represents the algorithm for the specific sampler. More...

class  cv::detail::tracking::TrackerContribSamplerCSC
TrackerSampler based on CSC (current state centered), used by MIL algorithm TrackerMIL. More...

class  cv::detail::tracking::TrackerFeature
Abstract base class for TrackerFeature that represents the feature. More...

class  cv::detail::tracking::TrackerFeatureFeature2d
TrackerContribFeature based on Feature2D. More...

class  cv::detail::tracking::TrackerFeatureHOG
TrackerContribFeature based on HOG. More...

class  cv::detail::tracking::TrackerFeatureLBP
TrackerContribFeature based on LBP. More...

class  cv::detail::tracking::TrackerFeatureSet
Class that manages the extraction and selection of features. More...

class  cv::detail::tracking::TrackerModel
Abstract class that represents the model of the target. More...

class  cv::detail::tracking::TrackerSampler
Class that manages the sampler in order to select regions for the update the model of the tracker [AAM] Sampling e Labeling. See table I and section III B. More...

class  cv::detail::tracking::TrackerSamplerAlgorithm
Abstract base class for TrackerSamplerAlgorithm that represents the algorithm for the specific sampler. More...

class  cv::detail::tracking::TrackerSamplerCS
TrackerContribSampler based on CS (current state), used by algorithm TrackerBoosting. More...

class  cv::detail::tracking::TrackerSamplerCSC
TrackerSampler based on CSC (current state centered), used by MIL algorithm TrackerMIL. More...

class  cv::detail::tracking::TrackerSamplerPF
This sampler is based on particle filtering. More...

class  cv::detail::tracking::TrackerStateEstimator
Abstract base class for TrackerStateEstimator that estimates the most likely target state. More...

class  cv::detail::tracking::TrackerStateEstimatorSVM
TrackerStateEstimator based on SVM. More...

class  cv::detail::tracking::TrackerTargetState
Abstract base class for TrackerTargetState that represents a possible state of the target. More...

## Typedefs

typedef std::vector< std::pair< Ptr< TrackerTargetState >, float > > cv::detail::tracking::ConfidenceMap
Represents the model of the target at frame $$k$$ (all states and scores) More...

typedef std::vector< Ptr< TrackerTargetState > > cv::detail::tracking::Trajectory
Represents the estimate states for all frames. More...

## Long-term optical tracking API

Long-term optical tracking is an important issue for many computer vision applications in real world scenario. The development in this area is very fragmented and this API is an unique interface useful for plug several algorithms and compare them. This work is partially based on [210] and [145] .

These algorithms start from a bounding box of the target and with their internal representation they avoid the drift during the tracking. These long-term trackers are able to evaluate online the quality of the location of the target in the new frame, without ground truth.

There are three main components: the TrackerContribSampler, the TrackerContribFeatureSet and the TrackerModel. The first component is the object that computes the patches over the frame based on the last target location. The TrackerContribFeatureSet is the class that manages the Features, is possible plug many kind of these (HAAR, HOG, LBP, Feature2D, etc). The last component is the internal representation of the target, it is the appearance model. It stores all state candidates and compute the trajectory (the most likely target states). The class TrackerTargetState represents a possible state of the target. The TrackerContribSampler and the TrackerContribFeatureSet are the visual representation of the target, instead the TrackerModel is the statistical model.

A recent benchmark between these algorithms can be found in [277]

If you want to create a new tracker, here's what you have to do. First, decide on the name of the class for the tracker (to meet the existing style, we suggest something with prefix "tracker", e.g. trackerMIL, trackerBoosting) – we shall refer to this choice as to "classname" in subsequent.

• Declare your tracker in modules/tracking/include/opencv2/tracking/tracker.hpp. Your tracker should inherit from Tracker (please, see the example below). You should declare the specialized Param structure, where you probably will want to put the data, needed to initialize your tracker. You should get something similar to :
class CV_EXPORTS_W TrackerMIL : public Tracker
{
public:
struct CV_EXPORTS Params
{
Params();
//parameters for sampler
int samplerInitMaxNegNum; // # negative samples to use during init
float samplerSearchWinSize; // size of search window
int samplerTrackMaxPosNum; // # positive samples to use during tracking
int samplerTrackMaxNegNum; // # negative samples to use during tracking
int featureSetNumFeatures; // #features
void read( const FileNode& fn );
void write( FileStorage& fs ) const;
};
of course, you can also add any additional methods of your choice. It should be pointed out, however, that it is not expected to have a constructor declared, as creation should be done via the corresponding create() method.
• Finally, you should implement the function with signature :
Ptr<classname> classname::create(const classname::Params &parameters){
...
}
That function can (and probably will) return a pointer to some derived class of "classname", which will probably have a real constructor.

Every tracker has three component TrackerContribSampler, TrackerContribFeatureSet and TrackerModel. The first two are instantiated from Tracker base class, instead the last component is abstract, so you must implement your TrackerModel.

### TrackerContribSampler

TrackerContribSampler is already instantiated, but you should define the sampling algorithm and add the classes (or single class) to TrackerContribSampler. You can choose one of the ready implementation as TrackerContribSamplerCSC or you can implement your sampling method, in this case the class must inherit TrackerContribSamplerAlgorithm. Fill the samplingImpl method that writes the result in "sample" output argument.

Example of creating specialized TrackerContribSamplerAlgorithm TrackerContribSamplerCSC : :

class CV_EXPORTS_W TrackerContribSamplerCSC : public TrackerContribSamplerAlgorithm
{
public:
TrackerContribSamplerCSC( const TrackerContribSamplerCSC::Params &parameters = TrackerContribSamplerCSC::Params() );
~TrackerContribSamplerCSC();
...
protected:
bool samplingImpl( const Mat& image, Rect boundingBox, std::vector<Mat>& sample );
...
};

Example of adding TrackerContribSamplerAlgorithm to TrackerContribSampler : :

//sampler is the TrackerContribSampler
Ptr<TrackerContribSamplerAlgorithm> CSCSampler = new TrackerContribSamplerCSC( CSCparameters );
return false;
//or add CSC sampler with default parameters
TrackerContribSamplerCSC, TrackerContribSamplerAlgorithm

### TrackerContribFeatureSet

TrackerContribFeatureSet is already instantiated (as first) , but you should define what kinds of features you'll use in your tracker. You can use multiple feature types, so you can add a ready implementation as TrackerContribFeatureHAAR in your TrackerContribFeatureSet or develop your own implementation. In this case, in the computeImpl method put the code that extract the features and in the selection method optionally put the code for the refinement and selection of the features.

Example of creating specialized TrackerFeature TrackerContribFeatureHAAR : :

class CV_EXPORTS_W TrackerContribFeatureHAAR : public TrackerFeature
{
public:
TrackerContribFeatureHAAR( const TrackerContribFeatureHAAR::Params &parameters = TrackerContribFeatureHAAR::Params() );
~TrackerContribFeatureHAAR();
void selection( Mat& response, int npoints );
...
protected:
bool computeImpl( const std::vector<Mat>& images, Mat& response );
...
};

Example of adding TrackerFeature to TrackerContribFeatureSet : :

//featureSet is the TrackerContribFeatureSet
Ptr<TrackerFeature> trackerFeature = new TrackerContribFeatureHAAR( HAARparameters );
TrackerContribFeatureHAAR, TrackerContribFeatureSet

### TrackerModel

TrackerModel is abstract, so in your implementation you must develop your TrackerModel that inherit from TrackerModel. Fill the method for the estimation of the state "modelEstimationImpl", that estimates the most likely target location, see [210] table I (ME) for further information. Fill "modelUpdateImpl" in order to update the model, see [210] table I (MU). In this class you can use the :cConfidenceMap and :cTrajectory to storing the model. The first represents the model on the all possible candidate states and the second represents the list of all estimated states.

Example of creating specialized TrackerModel TrackerMILModel : :

class TrackerMILModel : public TrackerModel
{
public:
TrackerMILModel( const Rect& boundingBox );
~TrackerMILModel();
...
protected:
void modelEstimationImpl( const std::vector<Mat>& responses );
void modelUpdateImpl();
...
};

bool TrackerMIL::initImpl( const Mat& image, const Rect2d& boundingBox )
{
...
//model is the general TrackerModel field of the general Tracker
model = new TrackerMILModel( boundingBox );
...
}

In the last step you should define the TrackerStateEstimator based on your implementation or you can use one of ready class as TrackerStateEstimatorMILBoosting. It represent the statistical part of the model that estimates the most likely target state.

Example of creating specialized TrackerStateEstimator TrackerStateEstimatorMILBoosting : :

class CV_EXPORTS_W TrackerStateEstimatorMILBoosting : public TrackerStateEstimator
{
class TrackerMILTargetState : public TrackerTargetState
{
...
};
public:
TrackerStateEstimatorMILBoosting( int nFeatures = 250 );
~TrackerStateEstimatorMILBoosting();
...
protected:
Ptr<TrackerTargetState> estimateImpl( const std::vector<ConfidenceMap>& confidenceMaps );
void updateImpl( std::vector<ConfidenceMap>& confidenceMaps );
...
};

//model is the TrackerModel of your Tracker
Ptr<TrackerStateEstimatorMILBoosting> stateEstimator = new TrackerStateEstimatorMILBoosting( params.featureSetNumFeatures );
model->setTrackerStateEstimator( stateEstimator );
TrackerModel, TrackerStateEstimatorMILBoosting, TrackerTargetState

During this step, you should define your TrackerTargetState based on your implementation. TrackerTargetState base class has only the bounding box (upper-left position, width and height), you can enrich it adding scale factor, target rotation, etc.

Example of creating specialized TrackerTargetState TrackerMILTargetState : :

class TrackerMILTargetState : public TrackerTargetState
{
public:
TrackerMILTargetState( const Point2f& position, int targetWidth, int targetHeight, bool foreground, const Mat& features );
~TrackerMILTargetState();
...
private:
bool isTarget;
Mat targetFeatures;
...
};

## ◆ ConfidenceMap

 typedef std::vector, float> > cv::detail::tracking::ConfidenceMap

#include <opencv2/video/detail/tracking.detail.hpp>

Represents the model of the target at frame $$k$$ (all states and scores)

See [210] The set of the pair $$\langle \hat{x}^{i}_{k}, C^{i}_{k} \rangle$$

TrackerTargetState

## ◆ Trajectory

 typedef std::vector > cv::detail::tracking::Trajectory

#include <opencv2/video/detail/tracking.detail.hpp>

Represents the estimate states for all frames.

[210] $$x_{k}$$ is the trajectory of the target up to time $$k$$