\n+
\n PARENTS = VL_AIB(PCX) runs Agglomerative Information Bottleneck\n (AIB) on the class-feature co-occurrence matrix PCX and returns a\n vector PARENTS representing the sequence of compressed AIB\n alphabets.\n
\n PCX is the joint probability of the occurrence of the class label\n C and the feature value X. PCX has one row for each class label\n"}, {"source1": "./usr/share/doc/libvlfeat-dev/doc/matlab/vl_aibhist.html", "source2": "./usr/share/doc/libvlfeat-dev/doc/matlab/vl_aibhist.html", "unified_diff": "@@ -62,15 +62,15 @@\n Documentation>MATLAB API>AIB - vl_aibhist\n
\n+
\n H = VL_AIBHIST(PARENTS, DATA) computes the histogram of the data\n points DATA on the VL_AIB tree defined by PARENTS. Each element of\n DATA indexes one of the leaves of the VL_AIB tree.\n
\n H = VL_AIBHIST(PARENTS, DATA, 'HIST') treats DATA as an histograms.\n In this case each compoment of DATA is the number of occurences of\n the VL_AIB leaves corresponding to that component.\n"}, {"source1": "./usr/share/doc/libvlfeat-dev/doc/matlab/vl_cf.html", "source2": "./usr/share/doc/libvlfeat-dev/doc/matlab/vl_cf.html", "unified_diff": "@@ -62,15 +62,15 @@\n Documentation>MATLAB API>PLOTOP - vl_cf\n
\n+
\n VL_COVDET() implements a number of co-variant feature detectors\n (e.g., DoG, Harris-Affine, Harris-Laplace) and corresponding\n feature descriptors (SIFT, raw patches).\n
\n F = VL_COVDET(I) detects upright scale and translation covariant\n features based on the Difference of Gaussian (Dog) cornerness\n measure from image I (a grayscale image of class SINGLE). Each\n"}, {"source1": "./usr/share/doc/libvlfeat-dev/doc/matlab/vl_ddgaussian.html", "source2": "./usr/share/doc/libvlfeat-dev/doc/matlab/vl_ddgaussian.html", "unified_diff": "@@ -62,15 +62,15 @@\n Documentation>MATLAB API>SPECIAL - vl_ddgaussian\n
\n+
\n Y=VL_DDGAUSSIAN(X) computes the second derivative of the standard\n Gaussian density.\n
\n To obtain the second derivative of the Gaussian density of\n standard deviation S, do\n
\n Y = 1/S^3 * VL_DDGAUSSIAN(X/S) .\n"}, {"source1": "./usr/share/doc/libvlfeat-dev/doc/matlab/vl_dwaffine.html", "source2": "./usr/share/doc/libvlfeat-dev/doc/matlab/vl_dwaffine.html", "unified_diff": "@@ -62,15 +62,15 @@\n Documentation>MATLAB API>IMOP - vl_dwaffine\n \n+
\n [DWX,DWY]=VL_DWAFFINE(X,Y) returns the derivative of the 2-D affine\n warp [WX; WY] = [A T] [X; Y] with respect to the parameters A,T\n computed at points X,Y.\n
\n See also: VL_WAFFINE(), VL_HELP().\n
\n+
\n MEMBERS=VL_ERFILL(I,ER) returns the list MEMBERS of the pixels which\n belongs to the extremal region represented by the pixel ER.\n
\n The selected region is the one that contains pixel ER and of\n intensity I(ER).\n
\n I must be of class UINT8 and ER must be a (scalar) index of the\n"}, {"source1": "./usr/share/doc/libvlfeat-dev/doc/matlab/vl_fisher.html", "source2": "./usr/share/doc/libvlfeat-dev/doc/matlab/vl_fisher.html", "unified_diff": "@@ -62,15 +62,15 @@\n Documentation>MATLAB API>FISHER - vl_fisher\n
\n+
\n ENC = VL_FISHER(X, MEANS, COVARIANCES, PRIORS) computes the Fisher\n vector encoding of the vectors X relative to the Gaussian mixture\n model with means MEANS, covariances COVARIANCES, and prior mode\n probabilities PRIORS.\n
\n X has one column per data vector (e.g. a SIFT descriptor), and\n MEANS and COVARIANCES one column per GMM component (covariance\n"}, {"source1": "./usr/share/doc/libvlfeat-dev/doc/matlab/vl_flatmap.html", "source2": "./usr/share/doc/libvlfeat-dev/doc/matlab/vl_flatmap.html", "unified_diff": "@@ -62,15 +62,15 @@\n Documentation>MATLAB API>QUICKSHIFT - vl_flatmap\n
\n+
\n [LABELS CLUSTERS] = VL_FLATMAP(MAP) labels each tree of the forest contained\n in MAP. LABELS contains the linear index of the root node in MAP, CLUSTERS\n instead contains a label between 1 and the number of clusters.\n
\n See also: VL_HELP().\n
\n+
\n [MEANS, COVARIANCES, PRIORS] = VL_GMM(X, NUMCLUSTERS) fits a GMM with\n NUMCLUSTERS components to the data X. Each column of X represent a\n sample point. X may be either SINGLE or DOUBLE. MEANS, COVARIANCES, and\n PRIORS are respectively the means, the diagonal covariances, and\n the prior probabilities of the Guassian modes. MEANS and COVARIANCES\n have the same number of rows as X and NUMCLUSTERS columns with one\n column per mode. PRIORS is a row vector with NUMCLUSTER entries\n"}, {"source1": "./usr/share/doc/libvlfeat-dev/doc/matlab/vl_hat.html", "source2": "./usr/share/doc/libvlfeat-dev/doc/matlab/vl_hat.html", "unified_diff": "@@ -62,15 +62,15 @@\n Documentation>MATLAB API>GEOMETRY - vl_hat\n
\n+
\n [TREE,ASGN] = VL_HIKMEANS(DATA,K,NLEAVES) applies integer K-menas\n recursively to cluster the data DATA, returing a structure TREE\n representing the clusters and a vector ASGN with the data to\n cluster assignments. The depth of the recursive partition is\n computed so that at least NLEAVES are generated.\n
\n VL_HIKMEANS() is built on top of VL_IKMEANS() and requires the\n"}, {"source1": "./usr/share/doc/libvlfeat-dev/doc/matlab/vl_kmeans.html", "source2": "./usr/share/doc/libvlfeat-dev/doc/matlab/vl_kmeans.html", "unified_diff": "@@ -62,15 +62,15 @@\n Documentation>MATLAB API>KMEANS - vl_kmeans\n
\n+
\n [C, A] = VL_KMEANS(X, NUMCENTERS) clusters the columns of the\n matrix X in NUMCENTERS centers C using k-means. X may be either\n SINGLE or DOUBLE. C has the same number of rows of X and NUMCENTER\n columns, with one column per center. A is a UINT32 row vector\n specifying the assignments of the data X to the NUMCENTER\n centers.\n
\n"}, {"source1": "./usr/share/doc/libvlfeat-dev/doc/matlab/vl_mser.html", "source2": "./usr/share/doc/libvlfeat-dev/doc/matlab/vl_mser.html", "unified_diff": "@@ -62,15 +62,15 @@\n Documentation>MATLAB API>MSER - vl_mser\n
\n+
\n R=VL_MSER(I) computes the Maximally Stable Extremal Regions (MSER)\n [1] of image I with stability threshold DELTA. I is any array of\n class UINT8. R is a vector of region seeds.\n
\n A (maximally stable) extremal region is just a connected component\n of one of the level sets of the image I. An extremal region can\n be recovered from a seed X as the connected component of the level\n"}, {"source1": "./usr/share/doc/libvlfeat-dev/doc/matlab/vl_quickvis.html", "source2": "./usr/share/doc/libvlfeat-dev/doc/matlab/vl_quickvis.html", "unified_diff": "@@ -62,15 +62,15 @@\n Documentation>MATLAB API>QUICKSHIFT - vl_quickvis\n
\n+
\n IEDGE = VL_QUICKVIS(I, RATIO, KERNELSIZE, MAXDIST, MAXCUTS) creates an edge\n stability image from a Quickshift segmentation. RATIO controls the tradeoff\n between color consistency and spatial consistency (See VL_QUICKSEG) and\n KERNELSIZE controls the bandwidth of the density estimator (See VL_QUICKSEG,\n VL_QUICKSHIFT). MAXDIST is the maximum distance between neighbors which\n increase the density.\n
\n"}, {"source1": "./usr/share/doc/libvlfeat-dev/doc/matlab/vl_rodr.html", "source2": "./usr/share/doc/libvlfeat-dev/doc/matlab/vl_rodr.html", "unified_diff": "@@ -62,15 +62,15 @@\n Documentation>MATLAB API>GEOMETRY - vl_rodr\n
\n+
\n R = VL_RODR(OM) where OM a 3-dimensional column vector computes the\n Rodrigues' formula of OM, returning the rotation matrix R =\n expm(vl_hat(OM)).\n
\n [R,DR] = VL_RODR(OM) computes also the derivative of the Rodrigues\n formula. In matrix notation this is the expression\n
\n"}, {"source1": "./usr/share/doc/libvlfeat-dev/doc/matlab/vl_setup.html", "source2": "./usr/share/doc/libvlfeat-dev/doc/matlab/vl_setup.html", "unified_diff": "@@ -62,15 +62,15 @@\n Documentation>MATLAB API>vl_setup\n \n+
\n PATH = VL_SETUP() adds the VLFeat Toolbox to MATLAB path and\n returns the path PATH to the VLFeat package.\n
\n VL_SETUP('NOPREFIX') adds aliases to each function that do not\n contain the VL_ prefix. For example, with this option it is\n possible to use SIFT() instead of VL_SIFT().\n
\n"}, {"source1": "./usr/share/doc/libvlfeat-dev/doc/matlab/vl_slic.html", "source2": "./usr/share/doc/libvlfeat-dev/doc/matlab/vl_slic.html", "unified_diff": "@@ -62,15 +62,15 @@\n Documentation>MATLAB API>SLIC - vl_slic\n
\n+
\n SEGMENTS = VL_SLIC(IM, REGIONSIZE, REGULARIZER) extracts the SLIC\n superpixes [1] from image IM. REGIONSIZE is the starting size of\n the superpixels and REGULARIZER is the trades-off appearance for\n spatial regularity when clustering (a larger value results in more\n spatial regularization). SEGMENTS is a UINT32 array containing the\n superpixel identifier for each image pixel.\n
\n"}, {"source1": "./usr/share/doc/libvlfeat-dev/doc/matlab/vl_tpfp.html", "source2": "./usr/share/doc/libvlfeat-dev/doc/matlab/vl_tpfp.html", "unified_diff": "@@ -62,15 +62,15 @@\n Documentation>MATLAB API>PLOTOP - vl_tpfp\n
\n+
\n [F,D] = VL_UBCREAD(FILE) reads the frames F and the descriptors D\n from FILE in UBC (Lowe's original implementation of SIFT) format\n and returns F and D as defined by VL_SIFT().\n
\n VL_UBCREAD(FILE, 'FORMAT', 'OXFORD') assumes the format used by\n Oxford VGG implementations .\n
\n"}, {"source1": "./usr/share/doc/libvlfeat-dev/doc/matlab/vl_vlad.html", "source2": "./usr/share/doc/libvlfeat-dev/doc/matlab/vl_vlad.html", "unified_diff": "@@ -62,15 +62,15 @@\n Documentation>MATLAB API>VLAD - vl_vlad\n
\n+
\n ENC = VL_VLAD(X, MEANS, ASSIGNMENTS) computes the VLAD\n encoding of the vectors X relative to cluster centers MEANS and\n vector-to-cluster soft assignments ASSIGNMENTS.\n
\n X has one column per data vector (e.g. a SIFT descriptor), and\n MEANS has one column per component. Usually one has one component\n per KMeans cluster and MEANS are the KMeans centers. X and MEANS\n"}, {"source1": "./usr/share/doc/libvlfeat-dev/doc/matlab/vl_xyz2rgb.html", "source2": "./usr/share/doc/libvlfeat-dev/doc/matlab/vl_xyz2rgb.html", "unified_diff": "@@ -62,15 +62,15 @@\n Documentation>MATLAB API>IMOP - vl_xyz2rgb\n
\n+
\n J = VL_XYZ2RGB(I) the XYZ image I in RGB format.\n
\n VL_XYZ2RGB(I,WS) uses the RGB workspace WS. WS is a string in\n
\n CIE: E illuminant and 2.2 gamma\n
\n Adobe: D65 illuminant and 2.2 gamma\n"}]}, {"source1": "xz --list", "source2": "xz --list", "unified_diff": "@@ -1,13 +1,13 @@\n Streams: 1\n Blocks: 1\n- Compressed size: 1676.8 KiB (1716996 B)\n+ Compressed size: 1676.7 KiB (1716964 B)\n Uncompressed size: 8170.0 KiB (8366080 B)\n Ratio: 0.205\n Check: CRC64\n Stream Padding: 0 B\n Streams:\n Stream Blocks CompOffset UncompOffset CompSize UncompSize Ratio Check Padding\n- 1 1 0 0 1716996 8366080 0.205 CRC64 0\n+ 1 1 0 0 1716964 8366080 0.205 CRC64 0\n Blocks:\n Stream Block CompOffset UncompOffset TotalSize UncompSize Ratio Check\n- 1 1 12 0 1716956 8366080 0.205 CRC64\n+ 1 1 12 0 1716924 8366080 0.205 CRC64\n"}]}]}]}