The Dataset class is an abstract wrapper describing a dataset. More...

#include <dataset.hpp>

Inheritance diagram for Dataset:

Public Member Functions
	Dataset (const std::string &base_location)
	Constructs a dataset given a base location. More...

	Dataset (const std::string &base_location, const std::string &db_data_location)
	Loads a dataset from the db_data_location. More...

virtual	~Dataset ()

virtual bool	write (const std::string &db_data_location)=0
	Writes the dataset mapping to the input data location. More...

virtual bool	read (const std::string &db_data_location)=0
	Reads the dataset mapping from the input data location. More...

virtual std::shared_ptr< Image >	image (uint64_t id) const =0
	Given a unique integer ID, returns an Image associated with that ID. More...

virtual uint64_t	num_images () const =0
	Returns the number of images in the dataset. More...

std::string	location () const
	Returns the absolute path of the data directory. More...

std::string	location (const std::string &relative_path) const
	Returns the absolute path of the file (appends the file path to the database path). More...

virtual bool	add_image (const std::shared_ptr< const Image > &image)=0
	Adds the given image to the database, if there is an id collision, will not add the image and return false, otherwise returns true. More...

std::vector< std::shared_ptr < const Image > >	all_images () const
	Returns a vector of all images in the dataset. More...

std::vector< std::shared_ptr < const Image > >	random_images (size_t count) const
	Returns a vector of random images in the dataset of size count. More...

std::vector< Dataset >	shard (const std::vector< std::string > &new_locations)
	: Shards the dataset to the new input locations, and returns the sharded datasets More...

Protected Attributes
std::string	data_directory

Detailed Description

The Dataset class is an abstract wrapper describing a dataset.

A dataset consiste of the actual data, plus a way to convert the images, or frames of a video into an integer index. The dataset should at minimum provide an easy way to map image paths to unique integers. For a sample implementation of a Dataset see the SimpleDataset class.

Combined with the Image class implementation, a Dataset + Image provides a way to find relevant paths for features and images. Note that the implementation of a Dataset or Image class should implement a relative path to the Image data, with the absolute path being interchangebale.

Definition at line 17 of file dataset.hpp.

Constructor & Destructor Documentation

Dataset::Dataset ( const std::string & base_location )

Constructs a dataset given a base location.

An example base location might be /c/data/. Given this base location, an implementation of the Dataset should find all the data and construct a mapping between the data and the id, for example by searching through base_location + /images/.

Definition at line 7 of file dataset.cxx.

References data_directory.

                                                {
         data_directory = base_location;
 }

Dataset::Dataset	(	const std::string &	base_location,
		const std::string &	db_data_location
	)

Loads a dataset from the db_data_location.

The base_location provides the absolute path of data.

Definition at line 11 of file dataset.cxx.

References data_directory.

                                                                                   {
         data_directory = base_location;
 }

Dataset::~Dataset ( )

virtual

Definition at line 15 of file dataset.cxx.

15 { }

Member Function Documentation

virtual bool Dataset::add_image ( const std::shared_ptr< const Image > & image )

pure virtual

Adds the given image to the database, if there is an id collision, will not add the image and return false, otherwise returns true.

Implemented in SimpleDataset.

std::vector< std::shared_ptr< const Image > > Dataset::all_images ( ) const

Returns a vector of all images in the dataset.

Definition at line 26 of file dataset.cxx.

References image(), and num_images().

Referenced by compute_bow(), compute_bow_features(), random_images(), and train_index().

                                                                   {
         std::vector< std::shared_ptr< const Image> > images(this->num_images());
         for (uint64_t i = 0; i < this->num_images(); i++) {
                 images[i] = this->image(i);
         }
         return images;
 }

virtual std::shared_ptr<Image> Dataset::image ( uint64_t id ) const

pure virtual

Given a unique integer ID, returns an Image associated with that ID.

Implemented in SimpleDataset.

Referenced by MatchesPage::add_match(), all_images(), benchmark_dataset(), compute_features(), InvertedIndex::search(), VocabTree::train(), and BagOfWords::train().

std::string Dataset::location ( ) const

Returns the absolute path of the data directory.

Definition at line 17 of file dataset.cxx.

References data_directory.

Referenced by MatchesPage::add_match(), bench_oxford(), benchmark_dataset(), compute_bow(), compute_bow_features(), compute_features(), main(), operator<<(), VocabTree::search(), InvertedIndex::search(), VocabTree::train(), BagOfWords::train(), InvertedIndex::train(), train_bow(), and train_tree().

                                   {
         return data_directory;
 }

std::string Dataset::location ( const std::string & relative_path ) const

Returns the absolute path of the file (appends the file path to the database path).

Definition at line 21 of file dataset.cxx.

References data_directory.

                                                                 {
         return data_directory + "/" + relative_path;
 }

virtual uint64_t Dataset::num_images ( ) const

pure virtual

Returns the number of images in the dataset.

Implemented in SimpleDataset.

Referenced by all_images(), benchmark_dataset(), compute_features(), operator<<(), and InvertedIndex::search().

std::vector< std::shared_ptr< const Image > > Dataset::random_images ( size_t count ) const

Returns a vector of random images in the dataset of size count.

Definition at line 34 of file dataset.cxx.

References all_images().

Referenced by compute_bow(), main(), train_bow(), and train_tree().

                                                                                  {
         std::vector< std::shared_ptr< const Image> > all = this->all_images();
         std::random_shuffle(all.begin(), all.end());
         std::vector< std::shared_ptr< const Image> > images(all.begin(), all.begin() + count);
         return images;
 }

virtual bool Dataset::read ( const std::string & db_data_location )

pure virtual

Reads the dataset mapping from the input data location.

Returns true if successful, false otherwise.

Implemented in SimpleDataset.

std::vector<Dataset> Dataset::shard ( const std::vector< std::string > & new_locations )

: Shards the dataset to the new input locations, and returns the sharded datasets

virtual bool Dataset::write ( const std::string & db_data_location )

pure virtual

Writes the dataset mapping to the input data location.

Returns true if successful, false otherwise.

Implemented in SimpleDataset.

Field Documentation

std::string Dataset::data_directory

protected

Definition at line 66 of file dataset.hpp.

Referenced by SimpleDataset::construct_dataset(), Dataset(), and location().

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

/home/psastras/vocabtree/modules/utils/dataset.hpp
/home/psastras/vocabtree/modules/utils/dataset.cxx

Public Member Functions

Protected Attributes

Detailed Description

Constructor & Destructor Documentation

Member Function Documentation

Field Documentation