1. You need to have istalled Openpose then you have to create a folder asset_openpose/ witch contains /bin,/python,/models and if you are on windows /x64 from Openpose.

2. You need to have in the folder json_in 2 files :

• One annotation file which look like :

• One json which contains the intrinsic parameters of the camera used :

3. You need to import your RGB and Depth images in the folders images\RGB and images\Depth. It is important that the images in RGB and in Depth file have the same names.

First of all delete all the files called ToDelete.txt in the empty folders.

Run the program called Main.py. You are going to get the first image of the your fastfood images.

Now you must select the desk with a simple Rois Selection. Draw your Rois selection and press enter to validate. Draw as much Rois as you want.

The more there is Rois the more the program is efficient. When you are done press esc. You'll get different lines :

Select the correct line of the end of the desk. You'll have to select 2 points of the line that are really on the desk.

Then the program is running. It can take a lot of time that depends on you computer configuration and the number of images imported.

You can see the results in the folder json_out. There will be 2 json File.

1. AnnotationWithArmsOverTheDesk.json is your Json file from the folder json_in where there is a new key :

The file AnnotationWithArmsOverTheDesk.json will be useful in https://github.com/cladr068/PlotCorrelationFrom3DDetection.

2. details.json is a recap for every images. We can see what point of open pose is above the desk. Warning : This file has nothing to do with the file details.json used in PlotCorrelation.