US 12,437,382 B2
Device and computer-implemented method for evaluating a control of a generator for determining pixels of a synthetic image
Edgar Schoenfeld, Tuebingen (DE); Anna Khoreva, Stuttgart (DE); and Julio Borges, Stuttgart (DE)
Assigned to ROBERT BOSCH GMBH, Stuttgart (DE)
Filed by Robert Bosch GmbH, Stuttgart (DE)
Filed on May 5, 2023, as Appl. No. 18/313,273.
Claims priority of application No. 22175795 (EP), filed on May 27, 2022.
Prior Publication US 2023/0386004 A1, Nov. 30, 2023
Int. Cl. G06V 20/70 (2022.01); G06T 7/00 (2017.01); G06T 11/00 (2006.01); G06V 10/74 (2022.01)
CPC G06T 7/0002 (2013.01) [G06T 11/00 (2013.01); G06V 10/761 (2022.01); G06V 20/70 (2022.01); G06T 2207/30168 (2013.01)] 9 Claims
OG exemplary drawing
 
1. A computer-implemented method for evaluating a control of a generator for determining pixels of a synthetic image, the generator being configured to determine pixels of the synthetic image from a first input including a label map and a first latent code, wherein the label map includes a mapping of at least one class to at least one of the pixels, the method comprising the following steps:
providing the label map and latent code, wherein the latent code includes input data points in a latent space;
providing the control, wherein the control includes a set of directions for moving the latent code in the latent space;
determining the first latent code depending on at least one input data point of the latent code that is moved in a first direction, wherein the first direction is selected from the set of directions; and
determining a distance between at least one pair of synthetic images, which are generated by the generator for different first inputs, wherein the different first inputs comprise the label map and vary by the first direction that is selected for determining the first latent code from the latent code;
wherein the first input includes a class mask, wherein the class mask indicates the input data points to be moved in the first direction for a class, and the distance is determined depending on pixels of the synthetic images indicated by the class mask to be considered.