ArgazkAI, smart system for organising photo collections:

Search engines and knowledge management

Overview

We have developed a system that allows the automatic generation of the descriptive metadata of photographs. The system automatically generates thematic tags (category, subcategory, keywords, objects and specific people) and descriptions of photographs using multimodal LLMs. These metadata are essential for getting the most out of the photograph catalogue. The more detailed the descriptive metadata, the more precise and efficient the searches are, ensuring maximised catalogue performance.

Challenge

The main aim of ArgazkAI is to create an intelligent system capable of generating metadata based on the photographs of photo agency FOKU, using algorithms and artificial intelligence techniques. In addition to automating the manual generation of metadata, this system allows the generation of more metadata and with greater detail, which significantly improves search efficiency.

Collaboration

It was developed in collaboration with the photo agency FOKU.

Result

The result of the project is an intelligent system capable of automatically analysing, describing and providing details about the photographs in FOKU’s catalogue. The system uses advanced artificial intelligence techniques —specifically, encoder language models and multimodal LLMs— to extract relevant information from the images and generate structured metadata. It can be installed in its entirety in a local environment, ensuring the control and security of the data. It is also a bilingual system, capable of understanding and generating content in Basque and in Spanish. 

Thanks to this system, the manual generation of metadata becomes automatic, providing a larger, more precise and consistent amount of information based on each photograph. It also significantly improves document management at FOKU, as it makes the search, organisation and recovery of large photo catalogues faster, more orderly and more efficient.