New Approaches to Archaeology using Deep Learning with Remote Sensor Data
Lead Research Organisation:
University of Southampton
Department Name: Electronics and Computer Science
Abstract
Deep learning (DL) for automated detection of archaeological sites (objects) on remote sensor (RS) data is a highly novel field. The key challenge of this field is in the inherent nature of the objects; they occur in small numbers, are sparsely located and feature a unique pattern on the different RS data modalities. To this extent we identify three main contributions, (1) to include multi-sensor data, (2) to optimise convolutional neural networks (CNNs) for small datasets and, (3) to optimise detection of the sparsely located objects.
Our preliminary results demonstrate that DL can be successfully applied to detect archaeological sites on each of the individual RS images, that our efforts to optimise CNNs for small datasets are promising, and that we need to further research a successful approach to integrate multi-modal data within CNNs. We aim to eventually create a workflow that is deemed useful to implement in archaeological research for detection of new archaeological sites.
Our preliminary results demonstrate that DL can be successfully applied to detect archaeological sites on each of the individual RS images, that our efforts to optimise CNNs for small datasets are promising, and that we need to further research a successful approach to integrate multi-modal data within CNNs. We aim to eventually create a workflow that is deemed useful to implement in archaeological research for detection of new archaeological sites.
Publications
Studentship Projects
Project Reference | Relationship | Related To | Start | End | Student Name |
---|---|---|---|---|---|
EP/N509747/1 | 01/10/2016 | 30/09/2021 | |||
1922247 | Studentship | EP/N509747/1 | 09/01/2017 | 08/01/2020 | Iris Kramer |
Description | See Engagement Activities section of researchfish. |
First Year Of Impact | 2018 |
Sector | Communities and Social Services/Policy,Education |
Impact Types | Societal |
Description | Collaboration on a case study on the Isle of Arran |
Organisation | Historic Environment Scotland |
Country | United Kingdom |
Sector | Public |
PI Contribution | We have shown the effectiveness of deep learning to detect archaeological sites on the Isle of Arran. With our approach, we discovered about 200 previously unknown sites. |
Collaborator Contribution | Historic Environment Scotland provided us with their LiDAR data and site locations required to train our models. After we did our bit we fed back our new detections to their team and they have confirmed that 200 of our total 300 detections were actual new sites. |
Impact | Outcomes: see the entry on our publication in British Archaeology. Disciplines involved: Archaeology, computer science (AI), earth observation. |
Start Year | 2019 |
Description | New Forest National Park |
Organisation | New Forest Park Authority |
Country | United Kingdom |
Sector | Private |
PI Contribution | The New Forest National Park authority archaeologist Lawrence Shaw was key to getting me known sites locations and LiDAR data used for automated detection of barrows in the Forest. |
Collaborator Contribution | They provided known sites locations and LiDAR data. |
Impact | Archaeology, deep learning, earth observation |
Start Year | 2017 |
Description | Sponsorship of PhD |
Organisation | Ordnance Survey |
Country | United Kingdom |
Sector | Public |
PI Contribution | My PhD stipend was paid by the Ordnance Survey and covered £14,296 pa subject to RCUK inflation rates. My fees were covered by UKRI. |
Collaborator Contribution | They provided me with aerial imagery that I used to make automated predictions for archaeological sites. |
Impact | Indirectly all outcomes can be attributed to this source. The provided data was not the main source for the successful detection outcome. |
Start Year | 2017 |
Description | Co-organiser of international workshop: MACHINE LEARNING IN ARCHAEOLOGY |
Form Of Engagement Activity | Participation in an activity, workshop or similar |
Part Of Official Scheme? | No |
Geographic Reach | International |
Primary Audience | Professional Practitioners |
Results and Impact | An international conference and workshop on 7-8 November 2019 in Rome, Italy, organised by the European Space Agency (ESA), the British School at Rome (BSR) and myself. Artificial Intelligence, Machine Learning and Deep Learning are opening new frontiers of inquiry. Application were presented on artefact analysis, text mining and remote sensing. Papers were presented at the BSR on November 7, followed by a workshop at ESA's European Space Research Institute (ESA/ESRIN) on November 8. |
Year(s) Of Engagement Activity | 2019 |
URL | http://blogs.esa.int/philab/2019/02/25/workshop-machine-learning-in-archaeology/ |
Description | Interdisciplinary Workshop on Artificial Intelligence' held on the 11th of May 2018, University of Southampton, United Kingdom |
Form Of Engagement Activity | Participation in an activity, workshop or similar |
Part Of Official Scheme? | No |
Geographic Reach | Local |
Primary Audience | Postgraduate students |
Results and Impact | I chaired and presented my research at a workshop on Artificial Intelligence that was held to inform our fellow postgraduate students of the opportunities that AI could have to their research. |
Year(s) Of Engagement Activity | 2018 |
URL | https://www.southampton.ac.uk/doctoral-college/research-community/festival-2018/machine-learning.pag... |
Description | Interview for international online news website TechRepublic. |
Form Of Engagement Activity | A press release, press conference or response to a media enquiry/interview |
Part Of Official Scheme? | No |
Geographic Reach | International |
Primary Audience | Media (as a channel to the public) |
Results and Impact | I was interviewed by a TechRepublic journalist to speak about practical applications of deep learning algorithms in the fields of archaeology and history. |
Year(s) Of Engagement Activity | 2018 |
URL | https://youtu.be/Io1lYck3xO8 |
Description | New Forest Knowledge Conference 2017: New Forest Historical Research and Archaeology: Who's doing it?, Lyndhurst, United Kingdom. 27 - 28 Oct 2017 |
Form Of Engagement Activity | A talk or presentation |
Part Of Official Scheme? | No |
Geographic Reach | Local |
Primary Audience | Public/other audiences |
Results and Impact | I provided a talk on my research at a local conference with topics relates to the New Forest. The video of my talk has now been reviewed over 400 times on Youtube - https://www.youtube.com/watch?v=UaVxJ8i-pBU&feature=youtu.be. |
Year(s) Of Engagement Activity | 2017 |
URL | https://nfknowledge.org/contributions/automated-detection-of-archaeology-in-the-new-forest-using-dee... |
Description | Publication in British Archaeology: "Scotland in miniature: automating archaeological survey on Arran" |
Form Of Engagement Activity | A magazine, newsletter or online publication |
Part Of Official Scheme? | No |
Geographic Reach | International |
Primary Audience | Public/other audiences |
Results and Impact | This piece was written by Dave Cowley from Historic Environment Scotland and myself on a case study we have collaborated on in Scotland. |
Year(s) Of Engagement Activity | 2019 |
URL | https://new.archaeologyuk.org/british-archaeology-magazine/ |