Why my 'bboxes' and 'scores' variables are all empty while certainly my image contains a person?

Hi,

This is my example image clearly with a person in it.

I would like to use Matlab's 'people detector' to detect the person in the image.

So I tried:

peopleDetector = vision.PeopleDetector;

I = imread('executive1.jpg');

[bboxes,scores] = step(peopleDetector,I);

I was hoping 'bboxes' could give me the position so that I could draw a box around the human in the image. However, 'bboxes' and 'scores' are all empty which I don't understand.

I'm a beginner to human detection. Please help.

Answer 1

[Alex Mcaulley]

Have you tried changing the peopledetector properties? For example, you get a result just changing the model:

peopleDetector = vision.PeopleDetector('ClassificationModel','UprightPeople_96x48');
I = imread('executive1.jpeg');
[bboxes,scores] = step(peopleDetector,I);
bboxes =

   103    66    73   125

Answer 2

[Matt J]

SVMs are allegedly poorer than neural net classifiers, but this helped,

I=imresize(I,[760,1064]);
peopleDetector = vision.PeopleDetector('UprightPeople_96x48','MinSize',[128,96]*2);