vlad.h File Reference

VLAD encoding (Vector of Locally Aggregated Descriptors (VLAD) encoding) More...

#include "generic.h"

Macros
VLAD options
#define	VL_VLAD_FLAG_NORMALIZE_COMPONENTS   (0x1 << 0)
	Normalize each VLAD component individually.
#define	VL_VLAD_FLAG_SQUARE_ROOT   (0x1 << 1)
	Use signed squared-root.
#define	VL_VLAD_FLAG_UNNORMALIZED   (0x1 << 2)
	Do not globally normalize the VLAD descriptor.
#define	VL_VLAD_FLAG_NORMALIZE_MASS   (0x1 << 3)
	Normalize each component by the number of features assigned to it.

Functions
void	vl_vlad_encode (void *enc, vl_type dataType, void const *means, vl_size dimension, vl_size numClusters, void const *data, vl_size numData, void const *assignments, int flags)
	VLAD encoding of a set of vectors.

---

Detailed Description

Author:

David Novotny

Andrea Vedaldi

See also:

Vector of Locally Aggregated Descriptors (VLAD) encoding

---

Function Documentation

void vl_vlad_encode	(	void *	enc,
		vl_type	dataType,
		void const *	means,
		vl_size	dimension,
		vl_size	numClusters,
		void const *	data,
		vl_size	numData,
		void const *	assignments,
		int	flags
	)

Parameters:

enc	output VLAD encoding (out).
dataType	the type of the input data (VL_TYPE_DOUBLE or VL_TYPE_FLOAT).
numData	number of data vectors to encode.
means	cluster means.
numClusters	number of clusters.
data	the data vectors to encode.
dimension	dimensionality of the data.
assignments	data to cluster soft assignments.
flags	options.

enc is the VLAD vector of size numClusters by dimension. means is a matrix with numClusters columns and dimension rows. data is the matrix of vectors to be encoded, with dimension rows and numData columns. assignments is a matrix with numClusters rows and numData columns. All the matrices should be stored in a row major order.

flag allows controlling further options: VL_VLAD_FLAG_NORMALIZE_COMPONENTS, VL_VLAD_FLAG_SQUARE_ROOT, VL_VLAD_FLAG_UNNORMALIZED, and VL_VLAD_FLAG_NORMALIZE_MASS.

See also:

Vector of Locally Aggregated Descriptors (VLAD) encoding