person-re-ranking/evaluation/LOMO_XQDA/code/Demo_LOMO.m

37 lines
832 B
Matlab

%% This is a demo for the LOMO feature extraction
clear; clc;
imgDir = '../images/';
addpath('../bin/');
%% Get image list
list = dir([imgDir, '*.bmp']);
n = length(list);
%% Allocate memory
info = imfinfo([imgDir, list(1).name]);
images = zeros(info.Height, info.Width, 3, n, 'uint8');
%% read images
for i = 1 : n
images(:,:,:,i) = imread([imgDir, list(i).name]);
end
%% extract features. Run with a set of images is usually faster than that one by one, but requires more memory.
descriptors = LOMO(images);
%% if you need to set different parameters other than the defaults, set them accordingly
%{
options.numScales = 3;
options.blockSize = 10;
options.blockStep = 5;
options.hsvBins = [8,8,8];
options.tau = 0.3;
options.R = [3, 5];
options.numPoints = 4;
descriptors = LOMO(images, options);
%}
rmpath('../bin/');