Back to results
US 12,608,867 B2Granted
System and method for generating a model of the human body
AssigneeSquaremind
InventorsTanguy Serrat, Ali Khachlouf
Life of this patent
Estimating the term…
Abstract
A method for generating a human body model includes acquiring at least one partial view of an individual defining a 3D image from at least one optical device arranged in a position of space, and generating a human body model by applying a trained neural network to generate a 3D model of the individual body from input 3D images, the training of the neural network being performed from a set of data describing partial views of the human body.
Claims21
A method for generating at least one human body model comprising:
- acquiring at least one partial view of at least one human body at least one individual the at least one partial view defining a 3D image from at least one optical device arranged in a position of space, and
- generating at least one 3D human body model of said at least one human body of the at least one individual from the 3D image by applying a trained neural network said neural network being trained to generate the at least one 3D human body model of said at least one human body of the at least one individual from at least one 3D input image of the at least one human body of the at least one individual, said training of the neural network being performed from a data set comprising a plurality of partial views of human bodies each representing a three-dimensional surface, each of the partial views used for training the neural network being generated from a three-dimensional mesh.
A method for generating at least one human body model comprising:
- acquiring at least one partial view of at least one individual defining a 3D image from at least one optical device arranged in a position of space, and
- generating at least one human body model by applying a trained neural network to generate at least one 3D model of said at least one body of the individual from at least one 3D input image, said training of the neural network being performed from a data set comprising a plurality of partial views of human bodies each representing a three-dimensional surface, each of the partial views used for training the neural network being generated from a three-dimensional mesh,
- wherein a human body parametric model is configured to generate a three-dimensional mesh of a human body, said three-dimensional mesh being used to generate a partial view of a human body in order to train the neural network;
- the human body parametric model being parameterizable from shape and joint coefficients defining a given configuration of a human body model.
1 more element
A method for generating at least one human body model comprising:
- acquiring at least one partial view of at least one individual defining a 3D image from at least one optical device arranged in a position of space, and
- generating at least one human body model by applying a trained neural network to generate at least one 3D model of said at least one body of the individual from at least one 3D input image, said training of the neural network being performed from a data set comprising a plurality of partial views of human bodies each representing a three-dimensional surface, each of the partial views used for training the neural network being generated from a three-dimensional mesh,
- wherein each partial view acquired defining a 3D image is encoded in the form of a point cloud to define inputs of the neural network,
- wherein the inputs of the neural network are defined by input vectors, each input vector comprising the coordinates of a point of the encoded point cloud and at least one piece of data encoding a local shape descriptor of the point cloud calculated at said point.
1 more element
Dependencies are inferred from the claim text.