Research · Ongoing
Research project in archaeology
DiDo: AI that detects stones in archaeological images
In the DiDo research project we are building an AI model that detects stones in archaeological images. After a feasibility test with a pre-trained model, we are training our own detection model, supported by AI:AT.
Starting point
Archaeological structures are still mostly recorded by hand from images. Experts review image after image, mark stones and document what they see. This takes a lot of time and depends heavily on the experience of individual people.
DiDo explores how far AI can take over and speed up this work. The goal is a model that reliably detects stones so experts can focus on interpretation rather than marking.
What we did
- Feasibility test with a pre-trained segmentation model: how well does AI detect stones without any training of its own? The test shows where general-purpose models reach their limits.
- Image data from the project team and the Austrian Archaeological Institute (ÖAI) collected and prepared for training.
- Our own object detection model trained and specialised in detecting stones in archaeological images.
- Fine-tuning and GPU compute with support from AI:AT, making research-grade training possible.
Outcome
The findings from the feasibility test feed directly into a model of our own, tailored to the needs of archaeology.
The project is ongoing. It shows how research, domain expertise and AI work together: archaeology provides the data and the understanding, we contribute the methods and the technical implementation.
More projects
Two tourism regions in Tyrol
Understanding guest behaviour from guest card data
Which guests come, which offers do they use and when, and how does their behaviour change? We analysed the guest card data of two regions so that strategic decisions rest on data.
Mid-sized company
Starting with AI, together with the whole team
A company is starting its AI journey and bringing every employee along. After a joint kick-off and clarifying the legal foundations, the focus is now on the right use cases.
Mountain railway in Austria
Who rides when? Usage analysis for a mountain railway
How is the mountain railway used, and by whom? From ticketing data we derived usage patterns, laid the groundwork for forecasting models and showed where marketing investment pays off.
Ready to turn your data into a competitive advantage?
Schedule a free 30-minute consultation. No commitment, just clarity.
30 minutes · free · no commitment · confidential