López de Hoyos Street 64, 1G. 28002 Madrid, Spain
Design innovation with vision models
Cerdá Group (Disney, main distributor)
GenAI, AI Vision & Tool Development
In the world of product design and labelling, precision and personalisation are essential, especially for a company with such high standards as Cerdá Group and Disney. At Logixs, we took on the challenge of transforming and automating key processes in the creation of creative templates and labelling using artificial intelligence.
Using the Florence 2 model, we optimised every step of the creative and production process, from the selection of Pantone colours to the generation of multilingual labels adapted to each market. Florence 2 demonstrated its ability to manage projects with a high degree of complexity, such as the creation of detailed dossiers for a wide range of products, including stationery, pet products and beauty accessories, integrating elements such as patterns, colours and specific licenses for well-known brands such as Disney, Marvel and Harry Potter. This ensures that each item maintains the brand identity, complies with international standards and facilitates its positioning in the global retail market. The AI model stood out as a key tool for mass customization.
Challenges
- Developing accurate AI for template recognition: Implementing the Florence 2 vision model to identify and manage specific areas on products and labels, ensuring accuracy in every detail of the product catalog.
- Color conversion between formats: Create a reliable tool that transforms HEX codes to Pantone, maintaining visual consistency across all products.
- Integration of different databases: Connect and synchronize multiple data sources (templates, products, labels) in a centralized system to ensure consistency and easy access to information.
- Managing High Volumes of Product Templates and Data: Develop a scalable structure that allows for storing and continuously updating a large number of templates, labels, and associated data.
Factors
Implementation of Advanced Vision AI
Centralized administrative panel
Scalability for expansion of functionalities
Process automation for operational efficiency
Efficient model training
Methodologies
Development of functional prototypes:
Creation of functional prototypes to validate each module of the system before its full integration. This allowed for fine-tuning of specific details in template recognition and colour conversion, ensuring that Cerdá standards were met.
Proof of Concept (PoC) Implementation
We developed and deployed a proof of concept (PoC) in a controlled environment, to validate the accuracy of the vision model.
Analysis and value proposition
Using a Design Thinking approach, we identified critical points in the design and labeling process, ensuring that each developed functionality provides direct value to the Cerdá team.
Feedback and adjustments with the client
Regular feedback sessions with the Cerdá team to incorporate their observations and make adjustments in real time, ensuring that the project adapts to the practical needs of their daily operation.
Scalability and preparation for future growth
Implementation of a scalable architecture that allows adding new functionalities and managing a greater volume of templates and data as the business grows, ensuring a long-term solution.
Results
"Implementing AI in design and labelling transformed Cerdá's processes, achieving a high degree of automation and accuracy. From centralised management of templates and resources, to automated colour conversion and advanced label recognition, each stage was optimised to maximise efficiency.
Thanks to this automation, production time was reduced by more than 70%, allowing for fast and accurate template generation. This efficiency not only increased the team's productivity, but also allowed Cerdá to focus on the innovation and quality of its products, strengthening its competitiveness in the market."