US 11,836,988 B2
Method and apparatus for recognizing an object from input data
Matthias Kirschner, Hildesheim (DE); and Thomas Wenzel, Hildesheim (DE)
Assigned to Robert Bosch GmbH, Stuttgart (DE)
Filed by Robert Bosch GmbH, Stuttgart (DE)
Filed on Aug. 5, 2021, as Appl. No. 17/394,872.
Claims priority of application No. 10 2020 209 979.0 (DE), filed on Aug. 6, 2020.
Prior Publication US 2022/0044035 A1, Feb. 10, 2022
Int. Cl. G06V 20/58 (2022.01); G06T 7/70 (2017.01); B60W 30/09 (2012.01); B60W 30/06 (2006.01)
CPC G06V 20/58 (2022.01) [B60W 30/06 (2013.01); B60W 30/09 (2013.01); G06T 7/70 (2017.01); B60W 2420/42 (2013.01); G06T 2207/20084 (2013.01)] 16 Claims
OG exemplary drawing
 
1. A method for recognizing an object from input data, the method comprising:
carrying out raw detections from the input data;
determining at least one first attribute for each raw detection, the at least one first attribute including a detection quality;
determining at least one second attribute for each raw detection;
determining a distance measure for at least one of the at least one first attribute and the at least one second attribute for each of the raw detections, the determined distance measure being temporally resolved or spatially resolved;
combining the raw detections into groups of raw detections based on a comparison of the determined distance measures to at least one defined distance measure;
identifying a selected group of the groups of raw detections as (i) a group of raw detections with a smallest distance measure in comparison with the distance measure of at least one other group of raw detections, or (ii) a group of raw detections having one raw detection formed by combining at least two of the raw detections, the combined raw detections having a smallest distance measure; and
determining the object to be recognized from the selected group of raw detections.