图像及label处理为密度图并保存为csv格式 固定核及自适应高斯核

图像及label处理为密度图并保存为csv格式

该方法使用matlab

427更新，解决了保存3通道图片，密度图3通道保存出错的问题

主程序

################################################
#############MERCY ROIN 2020/03/21##############
########https://blog.csdn.net/qq_42732137#######
################################################
clc; clear all;
seed = 95461354;
rng(seed)
N = 2;
dataset = 'train';
%图片链接
path = ['I:/crowd_dataset/L/'];
output_path = 'I:/crowd_dataset/L/formatted_trainval_1';
train_path_img = strcat(output_path, '/train/');
train_path_den = strcat(output_path, '/train_den/');
val_path_img = strcat(output_path, '/val/');
val_path_den = strcat(output_path, '/val_den/');
gt_path = ['I:/crowd_dataset/L/'];

img = dir('I:\crowd_dataset/L\*.jpg');%load img

mkdir(output_path)
mkdir(train_path_img);
mkdir(train_path_den);
mkdir(val_path_img);
mkdir(val_path_den);

num_images = length(img);
num_val = ceil(num_images*0.1);
indices = randperm(num_images);

for idx = 1:num_images
    i = indices(idx);
    imgname = img(idx).name;
    matname = imgname;
    csvname = imgname;
    matname(end-2:end) = 'mat';
    csvname(end-2:end) = 'csv';
    load([gt_path matname]);%gt.mat path
    
    if (mod(idx,10)==0)
        fprintf(1,'Processing %3d/%d files\n', idx, num_images);
    end

    image = imread([path, imgname]);
    annPoints = gt_point; %mat里函数名注意修改为自己的
    %灰度处理。如果图像使用RGB
%     [h, w, c] = size(image)