20240054810. LEARNING METHOD FOR A MACHINE LEARNING SYSTEM FOR DETECTING AND MODELING AN OBJECT IN AN IMAGE, CORRESPONDING COMPUTER PROGRAM PRODUCT AND DEVICE simplified abstract (FITTINGBOX)

From WikiPatents
Jump to navigation Jump to search

LEARNING METHOD FOR A MACHINE LEARNING SYSTEM FOR DETECTING AND MODELING AN OBJECT IN AN IMAGE, CORRESPONDING COMPUTER PROGRAM PRODUCT AND DEVICE

Organization Name

FITTINGBOX

Inventor(s)

Xavier Naturel of AUZEVILLE-TOLOSANE (FR)

Ariel Choukroun of SAINT ORENS DE GAMEVILLE (FR)

LEARNING METHOD FOR A MACHINE LEARNING SYSTEM FOR DETECTING AND MODELING AN OBJECT IN AN IMAGE, CORRESPONDING COMPUTER PROGRAM PRODUCT AND DEVICE - A simplified explanation of the abstract

This abstract first appeared for US patent application 20240054810 titled 'LEARNING METHOD FOR A MACHINE LEARNING SYSTEM FOR DETECTING AND MODELING AN OBJECT IN AN IMAGE, CORRESPONDING COMPUTER PROGRAM PRODUCT AND DEVICE

Simplified Explanation

The abstract describes a learning method for a machine learning system that involves obtaining augmented reality images, learning information, and delivering a set of parameters.

  • The method involves obtaining augmented reality images that include a real image and at least one virtual element representative of the object and/or the characteristic region.
  • For each augmented reality image, learning information is obtained, which includes a model for segmenting the given virtual element and a set of contour points corresponding to a parameterization of the given virtual element.
  • The machine learning system learns on the basis of the plurality of augmented reality images and the learning information, delivering a set of parameters that enable the system to detect and model the object and/or the characteristic region in a given image.

Potential applications of this technology:

  • Augmented reality applications: This method can be used in various augmented reality applications where virtual elements need to be accurately detected and modeled in real-time.
  • Object recognition: The machine learning system can be trained to detect and model specific objects or characteristic regions, making it useful for object recognition tasks in fields such as computer vision and robotics.
  • Virtual try-on: This technology can be applied in virtual try-on applications, where virtual elements representing clothing or accessories can be accurately placed on a real image of a person.

Problems solved by this technology:

  • Accurate detection and modeling of virtual elements: The method addresses the challenge of accurately detecting and modeling virtual elements in augmented reality images, improving the overall realism and effectiveness of augmented reality applications.
  • Object and characteristic region segmentation: The model for segmenting virtual elements helps in accurately separating them from the real image, enabling better understanding and manipulation of the augmented reality scene.

Benefits of this technology:

  • Improved augmented reality experiences: By accurately detecting and modeling virtual elements, this technology enhances the realism and immersion of augmented reality experiences.
  • Efficient object recognition: The machine learning system's ability to detect and model objects or characteristic regions can be leveraged for various applications, such as object tracking, scene understanding, and robotics.
  • Enhanced virtual try-on experiences: By accurately placing virtual elements on real images, this technology can provide more realistic and personalized virtual try-on experiences, benefiting industries like fashion and e-commerce.


Original Abstract Submitted

a learning method of a machine learning system carries out the steps of: obtaining augmented reality images including a real image and at least one virtual element representative of the object and/or the characteristic region; obtaining, for each augmented reality image, learning information including, for at least one given virtual element of the augmented reality image: a model for segmenting the given virtual element, and a set of contour points corresponding to a parameterisation of the given virtual element; and learning on the basis of the plurality of augmented reality images and the learning information, delivering a set of parameters enabling the machine learning system to detect and model the object and/or the characteristic region in a given image.

Retrieved from "http://wikipatents.org/index.php?title=20240054810._LEARNING_METHOD_FOR_A_MACHINE_LEARNING_SYSTEM_FOR_DETECTING_AND_MODELING_AN_OBJECT_IN_AN_IMAGE,_CORRESPONDING_COMPUTER_PROGRAM_PRODUCT_AND_DEVICE_simplified_abstract_(FITTINGBOX)&oldid=33698"