大型动物行为交互的时间统计1

本博客总结的是，统计狗行为交互阶段的时间所需的程序。统计时间分别是：亲近人的总时间，亲近饲养员休息的时间，亲近饲养员社交的时间； 亲近陌生人的总时间，亲近陌生人社交的时间，亲近陌生人休息的时间； 远离人的时间，远离人休息的时间，远离人社交的时间；

技术难点： 因为使用bonsai得到各段时间，使用像素来定义在哪一个区域（饲养员，陌生人，远离他们），因为场地背景是蓝色不好处理； 另外由于饲养员腿的位置造成干扰，统计时间造成不准确；

未来展望： 未来期望使用深度学习的方法，来识别狗的站立和蹲下的状态；站立定义为社交，蹲下定义为休息；以及利用坐标距离确定所在的区域；（这种方法比较麻烦，但是更加精确）；

 

将相关的bonsai代码和matlab代码粘贴如下：

% 本程序是用来分别统计在饲养员(CG)、陌生人(UF)、远离人区域的时间
clc;clear;
path = 'G:\行为学视频\狗_北京遗传所\2019_06_10\2019_06_10\';
name = '20190425635';
filename = '统计对比.xlsx'
filename2 = [path name filename];
zhenlv = 24.0;
time_interval_crop = 6.5;              %持续时间持续4s
time_interval_away = 2;                %狗在外部区域的持续时间
zhenshu_interval_crop  = zhenlv*time_interval_crop;                %持续的帧数
zhenshu_interval_away = zhenlv*time_interval_away;

filename = [name, '_region0.csv'];
filename1 = [name,'_walking0.csv'];
A = xlsread([path filename]);
a_speed = xlsread([path filename1]);
video_time = length(A)/zhenlv;
uf1_index = find(A(:,1) ==1);
uf1_time = length(uf1_index)/zhenlv;             %在陌生人所在区域时间           
cg_index =  find(A(:,1)==3);
cg_time = length(cg_index)/zhenlv;               %与饲养员所在区域的时间
away_index = find(A(:,1)==0);
away_time1 = length(away_index)/zhenlv;                  %远离人所在区域总时间
away_time = length(A)/zhenlv - uf1_time-cg_time;         %使用这个



T = find((away_index)>=length(a_speed));
away_index(T) =length(a_speed);
uf1_speed = (a_speed(uf1_index));   
uf1_rest_index = uf1_index(find(uf1_speed(:,1)==1));
uf1_diff = diff(uf1_rest_index);
uf1_diff_seris_count = series_count_accelerate(uf1_diff);
ufl_time_num = find(uf1_diff_seris_count(:,2)>zhenshu_interval_crop);
uf1_rest = sum(uf1_diff_seris_count(ufl_time_num,2))/zhenlv               %陌生人区域小狗休息的时间

uf1_social = uf1_time - uf1_rest;                              %陌生人与小狗社交的时间
cg_speed = (a_speed(cg_index));

cg_rest_index = cg_index(find(cg_speed(:,1)==1));
cg_diff = diff(cg_rest_index);
cg_diff_seris_count = series_count_accelerate(cg_diff);
cg_time_num = find(cg_diff_seris_count(:,2)>zhenshu_interval_crop);
cg_rest = sum(cg_diff_seris_count(cg_time_num,2))/zhenlv            %饲养员区域小狗休息时间

% cg_rest  =  length(find(cg_speed(:,1)==1))/zhenlv;  
cg_social = cg_time - cg_rest;                                      %与饲养员社交时间
away_speed = (a_speed(away_index));

away_rest_index = away_index(find(away_speed(:,1)==1));
away_diff = diff(away_rest_index);
away_diff_seris_count = series_count_accelerate(away_diff);
away_time_num = find(away_diff_seris_count(:,2)>zhenshu_interval_away);
away_rest = sum(away_diff_seris_count(away_time_num,2))/zhenlv

away_rest =  length(find(away_speed(:,1)==1))/zhenlv;
away_social =  away_time - away_rest;
time.toal = video_time;
time.uf1_time = uf1_time;
time.cg_time = cg_time;
time.away_time = away_time;
time.rest.uf1_rest = uf1_rest;
time.rest.cg_rest = cg_rest;
time.rest.away_rest = away_rest;
time.social.uf1_social =uf1_social;
time.social.cg_social = cg_social;
time.social.away_social = away_social;
t1 = {name,cg_time,cg_social,uf1_time,uf1_social,cg_time+uf1_time,cg_rest,uf1_rest,cg_rest+uf1_rest,away_time,away_rest,away_social,video_time};
xlswrite(filename2,t1,1,'F5:R5');