Define the problem
We identify the question, the sources and the criteria for evaluating results. Data preparation may involve images, texts, 3D models, geospatial data and organisational data.
Maindh_Lab · UNIVPM
AI algorithm and model development
From the research question to data, training and evaluation: Maindh_Lab designs and integrates AI systems with computing expertise and knowledge of their application contexts.
Our approach
Based at Università Politecnica delle Marche, the laboratory connects research, design and experimentation with institutions and companies.
Develop and train
We design algorithms and pipelines, train or adapt computer vision models and LLM-based systems, and integrate solutions into workflows and software.
Evaluate with people
We test the performance and limits of models. Domain experts interpret results and take part in decisions; human oversight remains integral to their use.
Where we apply these methods
Cultural heritage is the laboratory’s original focus; its methods now extend to other creative and productive fields.
Cultural heritage
Art analysis, searchable archives, 3D documentation and tools for cultural interpretation.
Fashion, design and creative industries
Digital and AI tools explored in support of creative and design processes.
Marine industry
AI and digital technologies applied to the sector’s data and processes.
Data and services
Preparing and analysing organisational data to inform decisions.
Research and projects
Explore the laboratory’s documented work: from AI for heritage and research projects to digital models and interactive experiences.