Daniel Canedo, João Fonte, Rita Dias, Tiago do Pereiro, Luís Gonçalves‐Seco, Marta Vázquez, Petia Georgieva, António J. R. Neves
{"title":"Automated Detection of Hillforts in Remote Sensing Imagery With Deep Multimodal Segmentation","authors":"Daniel Canedo, João Fonte, Rita Dias, Tiago do Pereiro, Luís Gonçalves‐Seco, Marta Vázquez, Petia Georgieva, António J. R. Neves","doi":"10.1002/arp.1958","DOIUrl":null,"url":null,"abstract":"Recent advancements in remote sensing and artificial intelligence can potentially revolutionize the automated detection of archaeological sites. However, the challenging task of interpreting remote sensing imagery combined with the intricate shapes of archaeological sites can hinder the performance of computer vision systems. This work presents a computer vision system trained for efficient hillfort detection in remote sensing imagery. Equipped with an adapted multimodal semantic segmentation model, the system integrates LiDAR‐derived LRM images and aerial orthoimages for feature fusion, generating a binary mask pinpointing detected hillforts. Post‐processing includes margin and area filters to remove edge inferences and smaller anomalies. The resulting inferences are subjected to hard positive and negative mining, where expert archaeologists classify them to populate the training data with new samples for retraining the segmentation model. As the computer vision system is far more likely to encounter background images during its search, the training data are intentionally biased towards negative examples. This approach aims to reduce the number of false positives, typically seen when applying machine learning solutions to remote sensing imagery. Northwest Iberia experiments witnessed a drastic reduction in false positives, from 5678 to 40 after a single hard positive and negative mining iteration, yielding a 99.3% reduction, with a resulting F<jats:sub>1</jats:sub> score of 66%. In England experiments, the system achieved a 59% F<jats:sub>1</jats:sub> score when fine‐tuned and deployed countrywide. Its scalability to diverse archaeological sites is demonstrated by successfully detecting hillforts and other types of enclosures despite their typical complex and varied shapes. Future work will explore archaeological predictive modelling to identify regions with higher archaeological potential to focus the search, addressing processing time challenges.","PeriodicalId":55490,"journal":{"name":"Archaeological Prospection","volume":null,"pages":null},"PeriodicalIF":2.1000,"publicationDate":"2024-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Archaeological Prospection","FirstCategoryId":"89","ListUrlMain":"https://doi.org/10.1002/arp.1958","RegionNum":3,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"0","JCRName":"ARCHAEOLOGY","Score":null,"Total":0}
引用次数: 0
Abstract
Recent advancements in remote sensing and artificial intelligence can potentially revolutionize the automated detection of archaeological sites. However, the challenging task of interpreting remote sensing imagery combined with the intricate shapes of archaeological sites can hinder the performance of computer vision systems. This work presents a computer vision system trained for efficient hillfort detection in remote sensing imagery. Equipped with an adapted multimodal semantic segmentation model, the system integrates LiDAR‐derived LRM images and aerial orthoimages for feature fusion, generating a binary mask pinpointing detected hillforts. Post‐processing includes margin and area filters to remove edge inferences and smaller anomalies. The resulting inferences are subjected to hard positive and negative mining, where expert archaeologists classify them to populate the training data with new samples for retraining the segmentation model. As the computer vision system is far more likely to encounter background images during its search, the training data are intentionally biased towards negative examples. This approach aims to reduce the number of false positives, typically seen when applying machine learning solutions to remote sensing imagery. Northwest Iberia experiments witnessed a drastic reduction in false positives, from 5678 to 40 after a single hard positive and negative mining iteration, yielding a 99.3% reduction, with a resulting F1 score of 66%. In England experiments, the system achieved a 59% F1 score when fine‐tuned and deployed countrywide. Its scalability to diverse archaeological sites is demonstrated by successfully detecting hillforts and other types of enclosures despite their typical complex and varied shapes. Future work will explore archaeological predictive modelling to identify regions with higher archaeological potential to focus the search, addressing processing time challenges.
期刊介绍:
The scope of the Journal will be international, covering urban, rural and marine environments and the full range of underlying geology.
The Journal will contain articles relating to the use of a wide range of propecting techniques, including remote sensing (airborne and satellite), geophysical (e.g. resistivity, magnetometry) and geochemical (e.g. organic markers, soil phosphate). Reports and field evaluations of new techniques will be welcomed.
Contributions will be encouraged on the application of relevant software, including G.I.S. analysis, to the data derived from prospection techniques and cartographic analysis of early maps.
Reports on integrated site evaluations and follow-up site investigations will be particularly encouraged.
The Journal will welcome contributions, in the form of short (field) reports, on the application of prospection techniques in support of comprehensive land-use studies.
The Journal will, as appropriate, contain book reviews, conference and meeting reviews, and software evaluation.
All papers will be subjected to peer review.