# US 2026/0094459 A1 · Method, Apparatus, Device, and Product for Annotating Images

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

- Number: US 2026/0094459 A1
- Document ID: US20260094459A1
- Status: Published application
- Kind: A1
- Application number: 19/342,515
- Filed: 2025-09-27
- Published: 2026-04-02
- Earliest claimed: 2024-09-30
- Assignees: Robert Bosch GmbH
- Applicants: Robert Bosch GmbH
- Inventors: Wenfu Wang; Xinrun Li
- Web page: https://patentgrep.com/patent/US20260094459A1

## CPC

- G06V 20/70: Labelling scene content, e.g. deriving syntactic or semantic representations
- G06F 40/247: Thesauruses; Synonyms
- G06T 7/70: Determining position or orientation of objects or cameras (camera calibration G06T7/80)
- G06V 10/26: Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
- G06V 10/82: using neural networks
- G06V 20/56: exterior to a vehicle by using sensors mounted on the vehicle
- G06T 2207/10028: Range image; Depth image; 3D point clouds
- G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL

## Abstract

A method, apparatus, device, and product for annotating images are disclosed. The method includes (i) determining existing annotation information, wherein the existing annotation information comprises content annotated by a user for images in a first subset of an image set, the image set comprising a plurality of images in the same domain, (ii) generating, by a language model, corresponding semantic content based on the existing annotation information, and (iii) annotating the images in a second subset of the image set based on the semantic content. In this way, the annotation information contained in the images in the image set that have been annotated by the user can be used to assist in the annotation task of the remaining images, thereby improving the efficiency and accuracy of annotation.

## Claims

### Claim 1

A method for annotating images, comprising:
acquiring existing annotation information, wherein the existing annotation information comprises content annotated by a user for images in a first subset of an image set, the image set comprising a plurality of images in the same domain;
generating, by a language model, corresponding semantic content based on the existing annotation information; and
annotating the images in a second subset of the image set based on the semantic content.

### Claim 2 (depends on 1)

The method according to claim 1, wherein generating, by a language model, corresponding semantic content comprises:
acquiring a plurality of category labels included in the existing annotation information;
generating, by the language model, content for interpreting the plurality of category labels; and
generating the semantic content based on the content for interpreting the plurality of category labels and the natural language used to describe image features corresponding to the images in the second subset.

### Claim 3 (depends on 2)

The method according to claim 2, wherein generating content for interpreting the plurality of category labels comprises:
generating content for interpreting the plurality of category labels based on a plurality of synonyms and/or a plurality of extensions of the plurality of category labels.

### Claim 4 (depends on 1)

The method according to claim 1, wherein annotating the images in a second subset of the image set comprises:
determining, based on a three-dimensional bounding box corresponding to an object contained in an image in the second subset, a two-dimensional box corresponding to the projected three-dimensional bounding box, the three-dimensional bounding box being a bounding box in the three-dimensional point cloud data corresponding to the image in the second subset; and
annotating the images in a second subset of the image set based on the semantic content and the two-dimensional box.

### Claim 5 (depends on 4)

The method according to claim 4, wherein determining the two-dimensional box corresponding to the projected three-dimensional bounding box comprises:
determining a plurality of two-dimensional position information corresponding to a plurality of corner points of the three-dimensional bounding box by projecting the three-dimensional bounding box onto an image plane;
determining a plurality of target corner points based on a plurality of two-dimensional position information corresponding to the plurality of corner points; and
determining a two-dimensional box corresponding to the three-dimensional bounding box based on the plurality of target corner points.

### Claim 6 (depends on 4)

The method according to claim 4, wherein annotating the images in a second subset of the image set comprises:
determining, based on the semantic content, two-dimensional candidate annotation boxes for objects contained in the images of the second subset; and
determining two-dimensional annotation boxes for the objects based on the two-dimensional candidate annotation boxes and the two-dimensional box.

### Claim 7 (depends on 6)

The method according to claim 6, wherein annotating the images in a second subset of the image set further comprises:
determining three-dimensional target point cloud data by sampling the three-dimensional point cloud data contained in the three-dimensional bounding box; and
determining the two-dimensional pixel points corresponding to the object by projecting the three-dimensional target point cloud data onto an image plane.

### Claim 8 (depends on 6)

The method according to claim 6, wherein the annotation content of the images in the second subset comprises the segmentation results corresponding to the images, and the method further comprises:
determining a semantic box for the object based on the segmentation result; and
determining an annotation quality score of the annotation content based on the geometric features and semantic features corresponding to the semantic box and the two-dimensional annotation box.

### Claim 9 (depends on 8)

The method according to claim 8, wherein the geometric features comprise whether the semantic box is continuous and the size of the two-dimensional annotation box, and determining the annotation quality score of the annotation content comprises:
determining a first annotation quality score of the annotation content based on whether the semantic boxes are continuous and whether the size of the two-dimensional annotation box matches the preset size of the object.

### Claim 10 (depends on 9)

The method according to claim 9, wherein determining the annotation quality score of the annotation content comprises:
determining a second annotation quality score of the annotation content according to the similarity between the semantic features and the semantic content.

### Claim 11 (depends on 10)

The method according to claim 10, wherein determining the annotation quality score of the annotation content comprises:
determining an annotation quality score of the annotation content based on the first annotation quality score and the second annotation quality score.

### Claim 12 (depends on 8)

The method according to claim 8, further comprising:
in response to the annotation quality score being less than a preset quality score threshold, re-annotating the image and determining the annotation quality score after re-annotation until the number of re-annotations is greater than a preset number threshold.

### Claim 13

An apparatus for generating an image, comprising:
an annotation information acquisition unit configured to acquire existing annotation information, wherein the existing annotation information comprises content annotated by a user for images in a first subset of an image set, the image set comprising a plurality of images in the same domain;
a semantic content generation unit configured to generate, by a language model, corresponding semantic content based on the existing annotation information; and
an annotation unit configured to annotate the images in a second subset of the image set based on the semantic content.

### Claim 14 (depends on 1)

An electronic device, comprising:
at least one processor; and
a memory coupled to the at least one processor and having instructions stored thereon, the instructions, when executed by the at least one processor, causing the device to perform the method according to claim 1.

### Claim 15 (depends on 1)

A computer program product, the computer program product being tangibly stored on a non-volatile computer-readable medium and comprising machine-executable instructions, the machine-executable instructions, when executed, causing a machine to execute steps of the method according to claim 1.
