Methods for generating and modifying synthetic on-person screening images

1 . A method for generating synthetic on-person screening images, comprising the steps of:

providing a database of labeled on-person screening images, wherein a labeling of a respective on-person screening image in the database indicates:

the presence of an object in the on-person screening image,

and, in case the object is present, at least one of: a classification of the object, a location of the object, a size of the object, a material of the object and a weight of the object;

training a generative artificial intelligence, AI, model by feeding at least some of the labeled on-person screening images in the database to the generative AI model;

receiving an input at an input interface, wherein the input comprises:

anatomic information of a person,

information on a presence of at least one further object, wherein the at least one further object is external to the person, and

information on a location of the further object relative to the person;

generating at least one synthetic on-person screening image with the generative AI model based on the input,

wherein the input comprises a script which comprises instructions for generating a plurality of synthetic on-person screening images for the at least one further object in different locations, orientations, shapes and/or sizes.

2 . The method of claim 1 ,

wherein the input further comprises: a classification, a concealment, a shape, a size, a material and/or a weight of the at least one further object.

3 . The method of claim 1 ,

wherein the input interface is configured to execute the script.

4 . The method of claim 1 ,

wherein the input interface comprises a user interface and the input is a manual user input.

5 . The method of claim 1 ,

wherein the input comprises at least one existing on-person screening image and information on how to modify said at least one existing on-person screening image.

6 . The method of claim 5 , further comprising the steps of:

virtualizing the at least one existing on-person screening image;

generating at least one synthetic image of the object using information from the at least one virtualized on-person screening image; and

modifying the at least one existing on-person screening image by inserting at least a section of the generated synthetic image(s) of the object into the respective existing image(s) of the person.

7 . The method of claim 6 ,

wherein the at least one synthetic image of the object is generated based on a posture of the person in the at least one virtualized image and at least one of: a location of the object relative to the person, a size of the object relative to the person, and a weight of the object.

8 . The method of claim 6 , further comprising the step of:

determining a shadow effect caused by the object on the body in the virtualized image;

wherein the step of modifying the at least one existing on-person screening image comprises adding the shadow effect to the at least one existing image.

9 . The method of claim 1 ,

wherein the input comprises a sketch or a drawing of the person with or without the at least one further object.

10 . The method of claim 1 ,

wherein the at least one synthetic on-person screening image is a synthetic MRI image, a synthetic mmWave image, a synthetic terahertz image or a synthetic nuclear resonance image.

11 . The method of claim 1 ,

wherein the generative AI model is implemented by a generative adversarial network and/or a variational autoencoder.

12 . The method of claim 1 ,

wherein the generative AI model is configured to execute a diffusion model and/or a large language model to generate the at least one synthetic on-person screening image.

13 . A method for training a machine learning algorithm for threat and/or posture detection in on-person screening images, comprising the steps of:

generating at least one synthetic on-person screening image using the method of claim 1 ;

feeding the at least one synthetic on-person screening image to the machine learning algorithm; and

training a threat and/or posture detection by the machine learning algorithm with the at least one synthetic on-person screening image.

14 . A system for generating synthetic on-person screening images, comprising:

a database of labeled on-person screening images, wherein a labeling of a respective on-person screening image in the database indicates:

the presence of an object in the on-person screening image,

and, in case the object is present, at least one of: a classification of the object, a location of the object, a size of the object, a material of the object and a weight of the object;

a generative artificial intelligence, AI, model; wherein, during a training routine, the generative AI model is configured to receive at least some of the labeled on-person screening images from the database;

an input interface configured to receive an input which comprises:

anatomic information of a person,

information on a presence of at least one further object, wherein the at least one further object is external to the person, and

information on a location of the further object relative to the person;

wherein the generative AI model is configured to generate at least one synthetic on-person screening image based on the input,

wherein the input comprises a script which comprises instructions for generating a plurality of synthetic on-person screening images for the at least one further object in different locations, orientations, shapes and/or sizes.

15 . The system of claim 14 ,

wherein the input interface comprises a user interface and/or a communication interface.