Recrié Co., Ltd. Initiates R&D of Image Analysis Algorithms Incorporating Expertise from Hiroshima University
Recrié Co., Ltd. has begun research and development of image and video analysis algorithms, leveraging the expertise of Hiroshima University, to combat risks like unauthorized use, alteration, and deepfakes in the era of generative AI, aiming to enhance content protection products.
📋 Article Processing Timeline
- 📰 Published: May 8, 2026 at 19:00
- 🔍 Collected: May 8, 2026 at 10:31
- 🤖 AI Analyzed: May 9, 2026 at 02:04 (15h 32m after Collected)
Recrié Co., Ltd. (Head Office: Hiroshima City, Hiroshima Prefecture; Representative Director: Yoshitaka Higaki) announces the commencement of research and development under the academic guidance of Associate Professor Toru Higaki of the Visual Informatics Laboratory, Graduate School of Advanced Science and Engineering, Hiroshima University. This initiative aims to advance product development related to similarity analysis of images and videos on the internet and the visualization of their spread.
This endeavor seeks to establish more sophisticated and practical analysis methods to address risks such as unauthorized use, alteration, and fake generation, in response to the rapidly changing environment surrounding corporate brand and creative assets.
1. Background
In recent years, with the evolution of generative AI and image editing technologies, the quantity and diversity of images and videos circulating on the internet have dramatically increased.
However,
- Design infringement by similar images
- Advertisements and fake content impersonating brands
- Difficulty distinguishing between legitimate and illegitimate secondary uses
These issues have become apparent, and conventional "identical image detection" is increasingly insufficient.
In response to this situation, Recrié is developing new products that combine in-depth analysis of "structural similarity" and "semantic similarity" with time-series visualization of content spread.
2. Initiatives
This initiative will involve research and development in the following areas under academic guidance:
- Selection of evaluation metrics and algorithms for image similarity judgment
- Design of similarity evaluation axes based on composition, shape, and semantic features
- Approaches for detecting fake images and videos
- Examination of extension methods for video data analysis
- Validation of evaluation axes in prototypes
Through this, we aim to establish a technological foundation that can quantitatively grasp not only detection but also "how it is similar" and "how it is spreading."
(*This initiative is based on Hiroshima University's academic guidance system and is not a joint research or co-authored publication with the university.)
■ Future Developments
Based on the technical knowledge established through this initiative, Recrié will accelerate the development of products that solve practical issues faced by companies in areas such as:
- Brand protection
- Content management
- Advertising monitoring
Furthermore, in the future, we envision systematizing research results and further enhancing them through industry-academia collaboration.
■ About Recrié Co., Ltd.
Recrié Co., Ltd. is a company that develops technologies to protect corporate brands and creative assets through the analysis of image and video content on the internet.
In recent years, with the evolution of generative AI and image editing technologies, issues such as unauthorized use, alteration, and increased similar expressions of content have become apparent. However, means to comprehensively and practically grasp these issues are not sufficiently developed.
To address this challenge, our company is working on building a mechanism that integrates technological areas such as:
- Analysis of structural and semantic similarity of images
- Detection of fake content
- Visualization of content spread on the internet
This mechanism will allow for a quantitative understanding of "what is being used and how."
Beyond being just a detection tool, we aim to establish a System (technological foundation) equipped with three elements—"Evidence," "Judgment," and "Intelligence"—to enable companies to make appropriate decisions. This will hereafter be referred to as EJIS (Aegis), and it will be constructed as a series of solutions.
Recrié will promote the development of products that support corporate decision-making and brand protection from both technical and practical aspects.
This endeavor seeks to establish more sophisticated and practical analysis methods to address risks such as unauthorized use, alteration, and fake generation, in response to the rapidly changing environment surrounding corporate brand and creative assets.
1. Background
In recent years, with the evolution of generative AI and image editing technologies, the quantity and diversity of images and videos circulating on the internet have dramatically increased.
However,
- Design infringement by similar images
- Advertisements and fake content impersonating brands
- Difficulty distinguishing between legitimate and illegitimate secondary uses
These issues have become apparent, and conventional "identical image detection" is increasingly insufficient.
In response to this situation, Recrié is developing new products that combine in-depth analysis of "structural similarity" and "semantic similarity" with time-series visualization of content spread.
2. Initiatives
This initiative will involve research and development in the following areas under academic guidance:
- Selection of evaluation metrics and algorithms for image similarity judgment
- Design of similarity evaluation axes based on composition, shape, and semantic features
- Approaches for detecting fake images and videos
- Examination of extension methods for video data analysis
- Validation of evaluation axes in prototypes
Through this, we aim to establish a technological foundation that can quantitatively grasp not only detection but also "how it is similar" and "how it is spreading."
(*This initiative is based on Hiroshima University's academic guidance system and is not a joint research or co-authored publication with the university.)
■ Future Developments
Based on the technical knowledge established through this initiative, Recrié will accelerate the development of products that solve practical issues faced by companies in areas such as:
- Brand protection
- Content management
- Advertising monitoring
Furthermore, in the future, we envision systematizing research results and further enhancing them through industry-academia collaboration.
■ About Recrié Co., Ltd.
Recrié Co., Ltd. is a company that develops technologies to protect corporate brands and creative assets through the analysis of image and video content on the internet.
In recent years, with the evolution of generative AI and image editing technologies, issues such as unauthorized use, alteration, and increased similar expressions of content have become apparent. However, means to comprehensively and practically grasp these issues are not sufficiently developed.
To address this challenge, our company is working on building a mechanism that integrates technological areas such as:
- Analysis of structural and semantic similarity of images
- Detection of fake content
- Visualization of content spread on the internet
This mechanism will allow for a quantitative understanding of "what is being used and how."
Beyond being just a detection tool, we aim to establish a System (technological foundation) equipped with three elements—"Evidence," "Judgment," and "Intelligence"—to enable companies to make appropriate decisions. This will hereafter be referred to as EJIS (Aegis), and it will be constructed as a series of solutions.
Recrié will promote the development of products that support corporate decision-making and brand protection from both technical and practical aspects.