| CPC G06F 18/285 (2023.01) [G06F 18/217 (2023.01); G06N 5/04 (2013.01); G06N 20/00 (2019.01); G06T 7/11 (2017.01); G06T 2207/20072 (2013.01); G06T 2207/20081 (2013.01); G06V 20/58 (2022.01); G06V 2201/06 (2022.01)] | 7 Claims |
|
1. A computer-implemented method for creating a machine learning system that is configured for segmentation and object detection in images, the machine learning system having one input for receiving an image and two outputs, a first output of the two outputs outputting the segmentation of the image and a second output of the two outputs outputting the object detection, the method comprising the following steps:
providing a directed graph, the directed graph having an input node, an output node, and a number of further nodes, the output node being connected via the further nodes using directed edges, and the nodes representing data and the edges representing operations that define a calculation rule and transfer a first node of the edges to further nodes connected to the respective edge;
selecting a first respective path through the graph, including:
from the number of further nodes, a subset is determined, all of whose nodes satisfy a predetermined characteristic with respect to data resolution,
from the subset, at least two additional nodes are selected,
the first selected respective path is a first path through the graph from the input node along the edges via a first one of the additional nodes up to the output node;
creating a first respective machine learning system as a function of the selected first respective path, wherein, in the creating of the first respective machine learning system step, those of the further nodes and directed edges that are on the selected first respective path are included in the first respective machine learning system, and those of the further nodes and directed edges that are not on the selected first respective path are excluded from the respective machine learning system;
training the created first respective machine learning system, and after the training of the created first respective machine learning system, adapted parameters of the first respective machine learning system being stored in corresponding edges of the directed graph;
selecting a second respective path through the graph, wherein the second selected respective path is a path through the graph from the input node along the edges via a second one of the additional nodes up to the output node, the second one of the additional nodes being different from the first one of the additional nodes, and wherein the selected second respective path is different from the selected first path;
creating a second respective machine learning system as a function of the selected second respective path, wherein, in the creating of the second respective machine learning system step, those of the further nodes and directed edges that are on the selected second respective path are included in the second respective machine learning system, and those of the further nodes and directed edges that are not on the selected second respective path are excluded from the respective machine learning system;
training the created second respective machine learning system, and after the training of the created second respective machine learning system, adapted parameters of the created second respective machine learning system being stored in corresponding edges of the directed graph;
and
after the training of the created first and second respective machine learning systems, creating the machine learning system as a function of the directed graph.
|