# US 2026/0094419 A1 · Device and method for training a machine learning system for feature extraction in digital images

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2026/0094419 A1
- Document ID: US20260094419A1
- Status: Published application
- Kind: A1
- Application number: 19/303,747
- Filed: 2025-08-19
- Published: 2026-04-02
- Earliest claimed: 2024-09-27
- Applicants: Robert Bosch GmbH
- Inventors: Levente Tamas; Victor Domsa
- Web page: https://patentgrep.com/patent/US20260094419A1

## CPC

- G06V 10/774: Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
- G06V 10/776: Validation; Performance evaluation
- G06V 10/82: using neural networks
- G06V 20/56: exterior to a vehicle by using sensors mounted on the vehicle
- G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING

## Abstract

A computer-implemented method of training a student machine learning system for feature extraction in digital images. The student machine learning system receives a digital image as input and provides different output data to the input digital image. Each of the different output data of the student machine learning system corresponds to the output data of one of at least two pretrained teacher machine learning systems for feature extraction in digital images. The teacher machine learning systems differ. The training includes receiving digital images as training data by the student machine learning system and by the at least two pretrained teacher machine learning systems, and training the student machine learning system by iteratively optimizing a loss function, which includes determining a deviation between the output data determined by the student machine learning system and the respective corresponding output data determined by the at least two teacher machine learning systems.

## Claims

### Claim 1

A computer-implemented method of training a student machine learning system for feature extraction from digital images, wherein at least two pretrained teacher machine learning systems for feature extraction from digital images are provided, wherein each teacher machine learning system is configured to receive a digital image as input and to provide respective output data indicating presence of at least one object and/or pattern in the digital image, wherein the teacher machine learning systems have different architectures, wherein the respective output data of the teacher machine learning systems differ from each other, wherein the student machine learning system is configured to receive a digital image as input and to provide different output data to said input digital image, wherein each one of the different output data of the student machine learning system corresponds to output data of one of the at least two teacher machine learning systems for the input digital image, the method comprising the following steps:
receiving digital images as training data by the student machine learning system and by the at least two pretrained teacher machine learning systems; and
training the student machine learning system by iteratively optimizing a loss function, wherein the loss function includes determining a deviation between output data determined by the student machine learning system and respective corresponding output data determined by the at least two teacher machine learning systems.

### Claim 2 (depends on 1)

The method according to claim 1, wherein: (i) the student machine learning system has an architecture differing from any of the teacher machine learning systems and/or (ii) the student machine learning systems has a reduced architecture as compared to at least one of the teacher machine learning systems, wherein the reduced architecture of the student machine learning system includes a reduced number of layers and/or connections between layers and/or a reduced number of parameters of the student machine learning system.

### Claim 3 (depends on 1)

The method according to claim 1, wherein the teacher machine learning systems have been trained using different training methods from each other, wherein a training method is defined by at least one of training data and/or loss function and/or training hyperparameters.

### Claim 4 (depends on 3)

The method according to claim 3, wherein the respective output data of the teacher machine learning systems differ in terms of at least one of semantic content and/or data representation and/or spatial resolution.

### Claim 5 (depends on 1)

The method according to claim 1, wherein the training data are obtained by providing a set of digital images, wherein each of the respective pre-trained teacher machine learning systems receives the digital images in the provided set as inputs, wherein the teacher machine learning systems determine a respective output to each of the provided input digital images, and wherein the training data for the student machine learning system are given by the set of digital images together with the respective corresponding outputs determined by each of the teacher machine learning systems.

### Claim 6 (depends on 1)

The method according to claim 1, wherein the loss function includes additional terms, wherein the additional terms include at least one of the loss functions used in the training of at least one of the teacher machine learning systems, respectively.

### Claim 7 (depends on 1)

The method according to claim 1, wherein the student machine learning system is further configured to determine a control signal for a vehicle by further processing the output data, and wherein the control signal is used to control an actuator.

### Claim 8

A training system, comprising:
a processor configured to train a student machine learning system for feature extraction from digital images,
wherein at least two pretrained teacher machine learning systems for feature extraction from digital images are provided, wherein each teacher machine learning system is configured to receive a digital image as input and to provide respective output data indicating presence of at least one object and/or pattern in the digital image, wherein the teacher machine learning systems have different architectures, wherein the respective output data of the teacher machine learning systems differ from each other, wherein the student machine learning system is configured to receive a digital image as input and to provide different output data to said input digital image, wherein each one of the different output data of the student machine learning system corresponds to output data of one of the at least two teacher machine learning systems for the input digital image, the processor being configured to:
receive digital images as training data by the student machine learning system and by the at least two pretrained teacher machine learning systems; and
train the student machine learning system by iteratively optimizing a loss function, wherein the loss function includes determining a deviation between output data determined by the student machine learning system and respective corresponding output data determined by the at least two teacher machine learning systems.

### Claim 9

A control system, configured to:
determine, using a trained student machine learning system, a control signal, wherein the control signal is used to control an actuator of the vehicle;
wherein the student machine learning system is trained for feature extraction from digital images, wherein at least two pretrained teacher machine learning systems for feature extraction from digital images are provided for the training, wherein each teacher machine learning system is configured to receive a digital image as input and to provide respective output data indicating presence of at least one object and/or pattern in the digital image, wherein the teacher machine learning systems have different architectures, wherein the respective output data of the teacher machine learning systems differ from each other, wherein the student machine learning system is configured to receive a digital image as input and to provide different output data to said input digital image, wherein each one of the different output data of the student machine learning system corresponds to output data of one of the at least two teacher machine learning systems for the input digital image, the student machine learning system is trained by:
receiving digital images as training data by the student machine learning system and by the at least two pretrained teacher machine learning systems; and
training the student machine learning system by iteratively optimizing a loss function, wherein the loss function includes determining a deviation between output data determined by the student machine learning system and respective corresponding output data determined by the at least two teacher machine learning systems.

### Claim 10

A non-transitory computer-readable storage medium on which is stored a computer program including instructions for training a student machine learning system for feature extraction from digital images, wherein at least two pretrained teacher machine learning systems for feature extraction from digital images are provided, wherein each teacher machine learning system is configured to receive a digital image as input and to provide respective output data indicating presence of at least one object and/or pattern in the digital image, wherein the teacher machine learning systems have different architectures, wherein the respective output data of the teacher machine learning systems differ from each other, wherein the student machine learning system is configured to receive a digital image as input and to provide different output data to said input digital image, wherein each one of the different output data of the student machine learning system corresponds to output data of one of the at least two teacher machine learning systems for the input digital image, the instructions, when execute by a computer, causing the computer to perform the following steps:
receiving digital images as training data by the student machine learning system and by the at least two pretrained teacher machine learning systems; and
training the student machine learning system by iteratively optimizing a loss function, wherein the loss function includes determining a deviation between output data determined by the student machine learning system and respective corresponding output data determined by the at least two teacher machine learning systems.
