ConvNet  1.0
A GPU-based C++ implementation of Convolutional Neural Nets
 All Classes Namespaces Functions Variables
Public Member Functions | Static Public Member Functions | Protected Member Functions | Protected Attributes | List of all members
DataIterator Class Referenceabstract

Base class for implementing data iterators. More...

#include <datahandler.h>

Inheritance diagram for DataIterator:
DummyDataIterator HDF5DataIterator< T > ImageDataIterator SlidingWindowDataIterator TextDataIterator

Public Member Functions

 DataIterator (const config::DataStreamConfig &config)
 
virtual void GetNext (float *data_out)=0
 
virtual void Seek (int row)
 
void Preprocess (Matrix &m)
 
void AddNoise (Matrix &input, Matrix &output)
 
int GetDims () const
 
int GetDataSetSize () const
 
void AddPCANoise (Matrix &m)
 
void SetJitterVariables (int max_offset)
 
void Jitter (Matrix &source, Matrix &dest)
 

Static Public Member Functions

static DataIteratorChooseDataIterator (const config::DataStreamConfig &config)
 

Protected Member Functions

void LoadMeans (const string &data_file)
 

Protected Attributes

int num_dims_
 
int dataset_size_
 
int row_
 
Matrix mean_
 
Matrix std_
 
Matrix pca_noise1_
 
Matrix pca_noise2_
 
Matrix eig_values_
 
Matrix eig_vectors_
 
Matrix width_offset_
 
Matrix height_offset_
 
Matrix flip_bit_
 
const string file_pattern_
 
const int num_colors_
 
const int gpu_id_
 
const bool translate_
 
const bool flip_
 
const bool normalize_
 
const bool pixelwise_normalize_
 
const bool add_pca_noise_
 
const float pca_noise_stddev_
 

Detailed Description

Base class for implementing data iterators.

Each data iterator handles a single stream of data. All derived classes must implement the GetNext() method and override the Seek() method appripriately.


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