# US 2025/0166402 A1 · Brand logo illegal use detection method, brand logo illegal use detection device and computer program

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

- Number: US 2025/0166402 A1
- Document ID: US20250166402A1
- Status: Published application
- Kind: A1
- Application number: 18/528,130
- Filed: 2023-12-04
- Published: 2025-05-22
- Earliest claimed: 2023-11-20
- Assignees: S2W Inc.
- Applicants: S2W Inc.
- Inventors: In Wook HWANG; Ki Cheol KIM; Keun Tae PARK
- Web page: https://patentgrep.com/patent/US20250166402A1

## CPC

- G06V 20/95: {Pattern authentication; Markers therefor; Forgery detection}
- G06Q 30/0185: {Product, service or business identity fraud}
- G06V 20/635: {Overlay text, e.g. embedded captions in a TV programme}
- G06V 30/19093: {Proximity measures, i.e. similarity or distance measures}
- G06V 30/19147: {Obtaining sets of training patterns; Bootstrap methods, e.g. bagging or boosting}
- G06V 2201/09: Recognition of logos
- G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06Q: INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR

## Abstract

The present disclosure relates to a method, device, and computer program capable of detecting whether a brand logo is illegally being used by an unauthorized person using an image and text of the brand logo. The present disclosure provides a detection technique capable of accurately detecting illegal use of a brand logo even when variously transformed forms and background images are used to avoid legal use detection. According to embodiments of the present disclosure, since the logo identification model can be generated through the machine learning dataset based on the logo transformation images and the image illegal use information can be generated using the logo identification model, it is possible to more accurately detect the illegal use of the logo that is used in variously transformed forms. In addition, since the illegal use information on the logo text can be considered in addition to the logo image, it is possible to further increase the accuracy of the illegal use detection.

## Claims

### Claim 1

A brand logo illegal use detection method comprising:
acquiring a logo image and logo text that are targets of illegal use detection;
generating image illegal use information indicating whether an illegal use suspect image in which the logo is suspected to be illegally used includes the logo image by using a logo identification model;
performing text analysis on the illegal use suspect image and generating text illegal use information indicating whether the logo text is illegally used; and
generating illegal use detection information on the illegal use suspect image based on the image illegal use information and the text illegal use information.

### Claim 2 (depends on 1)

The brand logo illegal use detection method of claim 1, further comprising:
prior to the generating of the image illegal use information, generating logo transformation images including transformed forms of the logo image to generate a machine learning dataset; and
performing learning based on the machine learning dataset to generate the logo identification model.

### Claim 3 (depends on 1)

The brand logo illegal use detection method of claim 1, wherein the generating of the machine learning dataset comprises:
generating the transformed forms based on at least one of rotation, transition, enlargement, reduction, proportion change, color change, brightness adjustment, transparency adjustment, and partial removal of the logo image; and
applying a plurality of background images to the transformed forms to generate the logo transformation images.

### Claim 4 (depends on 3)

The brand logo illegal use detection method of claim 3, wherein the generating of the machine learning dataset further comprises adding noise text and a noise image to the logo transformation images.

### Claim 5 (depends on 3)

The brand logo illegal use detection method of claim 3, wherein the illegal use suspect image includes a product description image uploaded to an online shopping mall and a phishing image uploaded to a phishing site.

### Claim 6 (depends on 3)

The brand logo illegal use detection method of claim 3, wherein the plurality of background images include a plurality of product description images uploaded to an online shopping mall and a plurality of public institution images used as a public institution description material.

### Claim 7 (depends on 1)

The brand logo illegal use detection method of claim 1, wherein the image illegal use information includes a probability that the illegal use suspect image includes the logo image, and the text illegal use information includes a probability that the illegal use suspect image includes the logo text.

### Claim 8 (depends on 7)

The brand logo illegal use detection method of claim 7, wherein the generating of the text illegal use information comprises:
performing optical character recognition (OCR) on the illegal use suspect image to extract illegal use suspect text; and
calculating a probability that the illegal use suspect image includes the logo text based on a similarity discrimination result between the illegal use suspect text and the logo text.

### Claim 9 (depends on 8)

The brand logo illegal use detection method of claim 8, wherein the generating of the text illegal use information comprises:
identifying surrounding text present around the illegal use suspect image and internal text present inside the illegal use suspect image in an online shopping mall or a phishing site; and
calculating a probability that the illegal use suspect image includes the logo text based on a similarity discrimination result between the surrounding text and the logo text and a similarity discrimination result between the internal text and the logo text.

### Claim 10 (depends on 9)

The brand logo illegal use detection method of claim 9, wherein the generating of the text illegal use information comprises:
querying texts associated with the logo text from an associated text pool of the brand logo; and
calculating a probability that the illegal use suspect image includes the logo text based on similarity discrimination results between each of the associated texts and the surrounding text and similarity discrimination results between each of the associated texts and the internal text.

### Claim 11 (depends on 8)

The brand logo illegal use detection method of claim 8, wherein the generating of the text illegal use information comprises:
when the illegal use suspect text is not extracted through the OCR, estimating that the illegal use suspect image includes a designed text; and
calculating a similarity between the designed text and the logo image using the logo identification model to calculate the probability that the illegal use suspect image includes the logo text.

### Claim 12

A non-transitory computer-readable recording medium on which a computer program to be executed by at least one processor is recorded, wherein the computer program includes instructions for:
analyzing a brand logo that is a target of illegal use detection to extract a logo image and logo text;
generating logo transformation images including transformed forms of the logo image to generate a machine learning dataset;
performing learning based on the machine learning dataset to generate a logo identification model;
generating image illegal use information indicating whether an illegal use suspect image in which the logo is suspected to be illegally used includes the logo image by using the logo identification model;
performing text analysis on the illegal use suspect image and generating text illegal use information indicating whether the logo text is illegally used; and
generating illegal use detection information on the illegal use suspect image based on the image illegal use information and the text illegal use information.

### Claim 13

A brand logo illegal use detection device comprising:
a memory configured to store instructions; and
a processor configured to execute the instructions to:
analyze a brand logo that is a target of illegal use detection to extract a logo image and logo text;
generate image illegal use information indicating whether an illegal use suspect image in which the logo is suspected to be illegally used includes the logo image by using a logo identification model;
perform text analysis on the illegal use suspect image and generating text illegal use information indicating whether the logo text is illegally used; and
generate illegal use detection information on the illegal use suspect image based on the image illegal use information and the text illegal use information.
