TRAIL 2022
Archaeological interpretation across scales – learning about landscapes using lidar
27 to 29 April 2022, Postojna, Slovenia
This international meeting builds on four previous meetings (TRAIL 2011, 2014, 2016 and 2018), and will be devoted to the study of airborne laser scanning (ALS) applications in archaeology, with a special focus on projects considering machine learning.
Open to researchers, PhD students, and professionals in archaeology and heritage management wishing to enhance their knowledge of lidar and who are interested in using machine learning for archaeological applications.
Presentations are in English.
In order to promote discussion and interactions, we encourage participants to prepare a poster on their own case study or project. We ask participants to submit a PDF draft of their poster 2 weeks before the event and bring the final printed version to the meeting.
TRAIL V Theme
Scale and detail are recurring themes in landscape archaeology and heritage management. They influence the data we use, the questions we ask, and the aims of projects. Archaeological projects are using lidar data in landscape work across a range of scales, from a single feature’s surroundings, to a valley, to a region. This TRAIL Meeting focuses on why and how we use lidar data at different scales, and the methodological challenges that arise as we address local and larger landscapes.
The use of Machine Learning (ML) to ‘scale up’ and work across extensive areas is now one of the most lively research topics in airborne lidar data use in archaeology. This interest is partly in response to advances made in Deep Learning, in particular Convolutional Neural Networks, across various disciplines. Applications of ML in remote sensing in general and in airborne lidar in particular – including automated object detection – already exhibit high performance and can, in some cases, match humans at challenging tasks. At the extensive landscape scale, ML is becoming an attractive alternative for traditional workflows based on visual interpretation of enhanced imagery.
At the local landscape scale, the questions and aims of research and management need different methods, and expert interpretation remains at the core of many projects. Obtaining more detailed data and complementary information remain real challenges, as does how to develop the expertise needed to interpret archaeological topography in its digital form. The challenges faced across these scales are connected, as expert interpretation of local landscapes feeds into the increasingly ML-supported methods used for extensive work.
At this TRAIL Meeting, we will ask: How can we build connections between the methods used at local and extensive scales? How can we improve data, visualisations, toolkits and training to support interpretative work at the local landscape scale? What parts of the workflow at an extensive scale can be replaced or improved by ML? How do we evaluate the results obtained by automated detection, and integrate the knowledge these processes create back into local-scale landscape interpretations?
These questions and themes will be addressed through a series of lectures and interactive workshops. The TRAIL Meeting will close with a roundtable to discuss key challenges and opportunities in the use of lidar for landscape interpretation across scales.