LogoAI
2024-05-30LogoAI • Automated generation of logos as vector graphics through the use of GANs
As part of the third round of the "Baden-Württemberg AI Innovation Competition" for the research and development of innovations in the field of artificial intelligence with a focus on "AI for SMEs" and "AI made in BW", d-serv GmbH from Tübingen started the research project "LogoAI - Automated generation of logos as vector graphics through the use of GANs" in late summer 2021. The aim of the AI Innovation Competition is for Baden-Württemberg to become an internationally leading location both in research and in the commercialisation of AI. The "LogoAI" concept presented was able to assert itself as an innovative project and won funding from the state of Baden-Württemberg.
The research application for the "LogoAI" research project was approved by the Baden-Württemberg Ministry of Economic Affairs, Labour and Tourism and the project began in August 2021 with a planned and adhered to duration until the end of 2022. For this research project, d-serv GmbH is cooperating with the Institute for Enterprise Systems (InES) at the University of Mannheim and Baden-Württemberg: Connected e.V. (bwcon). The aim of this cooperation is to develop an AI application for the automated creation of logos, signets or so-called key visuals in the form of vector graphics. Users should be able to enter their specific requirements, such as preferred colours, fonts and similar information, via a graphical user interface, on the basis of which the AI application then generates a new, unique and vector-based graphic with the help of a "Generative Adversarial Network" (GAN). Vector-based graphics, which have become the standard in modern 2D design, can be used within the web-to-publish solution ci-book™ developed by d-serv GmbH for the production of advertising and communication materials.
Further information from the Ministry can be found on the following website - In German: Innovation competition "AI for SMEs": Ministry of Economic Affairs funds 36 AI projects with a total of 7.5 million euros
The news from cooperation partner InES can be found on the following website - In German: LogoAI research project launches with financial support from the Baden-Württemberg Ministry of Economic Affairs, Labour and Tourism
d-serv GmbH would like to thank the Baden-Württemberg Ministry of Economic Affairs, Labour and Tourism for its successful support in implementing the project and, above all, its cooperation partners for their excellent and imaginative collaboration.
Result: ci-book™ • Key Visual Creator
d-serv GmbH and the Institute for Enterprise Systems (InES) at the University of Mannheim are delighted to present the result of their successful collaboration at the start of 2023. The LogoAI project goes back to the Baden-Württemberg AI innovation competition announced in late summer 2021 as part of the initiative of the Ministry of Economic Affairs, Labour and Housing in Baden-Württemberg, which promotes the use of AI in SMEs with a total of 10 million euros.
With the ci-book™ • Key Visual Creator, which has been available as an option since the beginning of 2023, individual specifications such as initial image, desired font and possible text information are first recorded in order to then generate an icon-text combination as a unique key visual in the form of a vector graphic with the help of LogoAI. The ci-book™ • Key Visual Creator thus offers a helpful and user-friendly application for numerous users with different levels of knowledge in the areas of design and layout, making it possible to create vector-based graphics in a simple, iterative process and integrate them directly into various advertising and communication media as required.
Procedure in the ci-book™ • Key Visual Creator
- The source image - for which the copyright or usage rights must be available - should ideally be available and imported in PNG format. This source image is then processed with the help of LogoAI using the "image-to-icon style" method in order to obtain a vector-based graphic as the basis for the key visual.
- After selecting a licence-free font or uploading your own licensed font, the text components of the key visual can be defined; for example, the company, brand or product name with or without an additional line of text, such as a subline or a claim.
- These individual components for image and text generated by LogoAI are then combined and can be edited to a limited extent (colours, dimensions). If the result is not satisfactory, the process and transmission of the data to LogoAI can be restarted.
- The generated key visual is available as a vector graphic and can be provided or exported as a PNG or SVG file format.
How LogoAI works
At the beginning of the process, you can import your own photo, which is then transformed by an AI in several steps into an abstracted signet for the new key visual. This photo is first segmented into several instances by an AI. In the next step, the objects resulting from the segmentation are transferred into various proposals by an AI in the so-called "image-to-icon style".
At the same time, several fonts are suggested based on previously defined brand and font attributes such as modern, classic, extravagant, playful or handwritten. The suggested or selectable fonts are taken from a database of classified and freely available fonts, also created using AI. The key visual and font are then combined to create a homogeneous logo. For this purpose, an AI creates the so-called "saliency map" for the respective key visual. The saliency map is a kind of heat map that shows which areas in an image are particularly striking or are perceived first by the human eye. The additionally created frames show possible positioning of lettering, from which the preferred variant can be selected.
LogoAI is able to create a fully-fledged logo that can be used as part of the corporate design for companies, products, services and corresponding marketing and communication measures. LogoAI takes on a core task in the design process, allowing users to act independently at numerous points in the design and creation of a logo; be it through the selection of the photo or basic image for the AI-generated key visual, through the definition of brand and font attributes or in the final selection of a suggested variant.