Submit a preprint

Latest recommendationsrsstwitter

IdTitleAuthorsAbstractPicture▲Thematic fieldsRecommenderReviewersSubmission date
02 May 2024
article picture

Machine Learning for UAV and Ground-Captured Imagery: Toward Standard Practices

A step forward in detecting small objects in UAV data for archaeological surveying

Recommended by ORCID_LOGO based on reviews by 2 anonymous reviewers

In this paper [1], the authors describe how they apply machine learning with YOLOv5 to classify visual data, aiming to enhance understanding of archaeological phenomena before conducting destructive fieldwork. Despite challenges, the integration of machine learning with remote sensing technology was seen as transformative, enabling precise recording of areas of interest and assessment of environmental risk factors. The paper discusses successes, failures, and future directions in machine learning research, emphasising the need for standardisation and integration of streamlined methods. The application of machine learning techniques facilitates non-destructive analysis of material culture records, improving conservation efforts and offering insights into both past and contemporary phenomena. While the initial use of YOLOv5 showed potential for consistent detection of archaeological features, further refinement and dataset enlargement are deemed necessary for broader application in non-destructive archaeological surveying. The authors advocate for the integration of machine learning tools in archaeological research to save time, resources, and promote ethical digital recording practices. They highlight the importance of standardised methodologies to enhance credibility and reproducibility, aiming to contribute to the ongoing dialogue in computational archaeology.

Overall, I think this paper is a good step forward in detecting small objects in UAV data, and contains useful information for similar studies. The aim towards greater reproducibility and standardisation is of course shared more widely in the machine learning community, and this study is a good example of how to approach this.

References

[1] Sharp, K., Christofis, B., Eslamiat, H., Nepal, U. and Osores Mendives, C. (2024). Machine Learning for UAV and Ground-Captured Imagery: Toward Standard Practices. Zenodo, 8307612, ver. 5 peer-reviewed and recommended by Peer Community in Archaeology. https://doi.org/10.5281/zenodo.8307612

Machine Learning for UAV and Ground-Captured Imagery: Toward Standard PracticesSharp Kayeleigh, Christofis Brooklyn, Eslamiat Hossein, Nepal Upesh, Osores Mendives Carlos<p>Our collaborative work began in 2019 with the intent to overcome obstacles that had arisen from the inability to access curated artifact collections from remote locations. It was our specific aim to not only create digital twins of excavated ob...Ceramics, Computational archaeology, Remote sensing, South AmericaAlex Brandsen2023-09-01 09:56:18 View
29 Jan 2024
article picture

Visual encoding of a 3D virtual reconstruction's scientific justification: feedback from a proof-of-concept research

3D Models, Knowledge and Visualization: a prototype for 3D virtual models according to plausible criteria

Recommended by based on reviews by Robert Bischoff and Louise Tharandt

The construction of 3D realities is deeply embedded in archaeological practices. From sites to artifacts, archaeology has dedicated itself to creating digital copies for the most varied purposes. The paper “Visual encoding of a 3D virtual reconstruction's 3 scientific justification: feedback from a proof-of-concept research” (Jean-Yves et al 2024) represents an advance, in the sense that it does not just deal with a three-dimensional theory for archaeological practice, but rather offers proposals regarding the epistemic component, how it is possible to represent knowledge through the workflow of 3D virtual reconstructions themselves. The authors aim to unite three main axes - knowledge modeling, visual encoding and 3D content reuse - (Jean-Yves et al 2024: 2), which, for all intents and purposes, form the basis of this article. With regard to the first aspect, this work questions how it is possible to transmit the knowledge we want to a 3D model and how we can optimize this epistemic component. A methodology based on plausibility criteria is offered, which, for the archaeological field, offers relevant space for reflection. Given our inability to fully understand the object or site that is the subject of the 3D representation, whether in space or time, building a method based on probabilistic categories is probably one of the most realistic approaches to the realities of the past.

Thus, establishing a plausibility criterion allows the user to question the knowledge that is transmitted through the representation, and can corroborate or refute it in future situations. This is because the role of reusing these models is of great interest to the authors, a perfectly justifiable sentiment, as it encourages a critical view of scientific practices. Visual encoding is, in terms of its conjunction with knowledge practices, a key element. The notion of simplicity under Maeda's (2006) design principles not only represents a way of thinking that favors operability, but also a user-friendly design in the prototype that the authors have created. This is also visible when it comes to the reuse of parts of the models, in a chronological logic: adapting the models based on architectural elements that can be removed or molded is a testament to intelligent design, whereby instead of redoing models in their entirety, they are partially used for other purposes.

All these factors come together in the final prototype, a web application that combines relational databases (RDBMS) with a data mapper (MassiveJS), using the PHP programming language.  The example used is the Marmoutier Abbey hostelry, a centuries-old building which, according to the sources presented, has evolved architecturally over several centuries ((Jean-Yves et al 2024: 8). These states of the building are represented visually through architectural elements based on their existence, location, shape and size, always in terms of what is presented as being plausible. This allows not only the creation of a matrix in which various categories are related to various architectural elements, but also a visual aid, through a chromatic spectrum, of the plausibility that the authors are aiming for. 

In short, this is an article that seeks to rethink the degree of knowledge we can obtain through 3D visualizations and that does not take models as static, but rather realities that must be explored, recycled and reinterpreted in the light of different data, users and future research. For this reason, it is a work of great relevance to theoretical advances in 3D modeling adapted to archaeology.

 

References

Blaise, J.-Y., Dudek, I., Bergerot, L. and Gaël, S. (2024). Visual encoding of a 3D virtual reconstruction's scientific justification: feedback from a proof-of-concept research, Zenodo, 7983163, ver. 3 peer-reviewed and recommended by Peer Community in Archaeology. https://doi.org/10.5281/zenodo.10496540

John Maeda. (2006). The Laws of Simplicity. MIT Press, Cambridge, MA, USA.

Visual encoding of a 3D virtual reconstruction's scientific justification: feedback from a proof-of-concept researchJ.Y Blaise, I.Dudek, L.Bergerot, G.Simon<p>&nbsp;3D virtual reconstructions have become over the last decades a classical mean to communicate &nbsp;about analysts’ visions concerning past stages of development of an edifice or a site. However, they still today remain quite often a one-s...Computational archaeology, Spatial analysisDaniel Carvalho2023-05-30 00:43:03 View
05 Jun 2023
article picture

SEAHORS: Spatial Exploration of ArcHaeological Objects in R Shiny

Analyzing piece-plotted artifacts just got simpler: A good solution to the wrong problem?

Recommended by based on reviews by Frédéric Santos, Jacqueline Meier and Maayan Lev

Paleolithic archaeologists habitually measure 3-coordinate data for artifacts in their excavations. This was first done manually, and in the last three decades it is usually performed by a total station and associated hardware. While the field recording procedure is quite straightforward, visualizing and analyzing the data are not, often requiring specialized proprietary software or coding expertise. Here, Royer and colleagues (2023) present the SEAHORS application, an elegant solution for the post-excavation analysis of artifact coordinate data that seems to be instantly useful for numerous archaeologists. SEAHORS allows one to import and organize field data (Cartesian coordinates and point description), which often comes in a variety of formats, and to create various density and distribution plots. It is specifically adapted to the needs of archaeologists, is free and accessible, and much simpler to use than many commercial programs. The authors further demonstrate the use of the application in the post-excavation analysis of the Cassenade Paleolithic site (see also Discamps et al., 2019). 

While in no way detracting from my appreciation of Royer et al.’s (2023) work, I would like to play the devil’s advocate by asking whether, in the majority of cases, field recording of artifacts in three coordinates is warranted. Royer et al. (2023) regard piece plotting as “…indispensable to propose reliable spatial planimetrical and stratigraphical interpretations” but this assertion does not hold in all (or most) cases, where careful stratigraphic excavation employing thin volumetric units would do just as well.

Moreover, piece-plotting has some serious drawbacks. The recording often slows excavations considerably, beyond what is needed for carefully exposing and documenting the artifacts in their contexts, resulting in smaller horizontal and vertical exposures (e.g., Gilead, 2002). This typically hinders a fuller stratigraphic and contextual understanding of the excavated levels and features. Even worse, the method almost always creates a biased sample of “coordinated artifacts”, in which the most important items for understanding spatial patterns and site-formation processes – the small ones – are underrepresented. Some projects run the danger of treating the coordinated artifacts as bearing more significance than the sieve-recovered items, preferentially studying the former with no real justification. Finally, the coordinated items often go unassigned to a volumetric unit, effectively disconnecting them from other types of data found in the same depositional contexts.  

The advantages of piece-plotting may, in some cases, offset the disadvantages. But what I find missing in the general discourse (certainly not in the recommended preprint) is the “theory” behind the seemingly technical act of 3-coordinate recording (Yeshurun, 2022). Being in effect a form of sampling, this practice needs a rethink about where and how to be applied; what depositional contexts justify it, and what the goals are. These questions should determine if all “visible” artifacts are plotted, or just an explicitly defined sample of them (e.g., elongated items above a certain length threshold, which should be more reliable for fabric analysis), or whether the circumstances do not actually justify it. In the latter case, researchers sometimes opt for using “virtual coordinates” within in each spatial unit (typically 0.5x0.5 m), essentially replicating the data that is generated by “real” coordinates and integrating the sieve-recovered items as well. In either case, Royer et al.’s (2023) solution for plotting and visualizing labeled points within intra-site space would indeed be an important addition to the archaeologists’ tool kits.

 

References cited 

Discamps, E., Bachellerie, F., Baillet, M. and Sitzia, L. (2019). The use of spatial taphonomy for interpreting Pleistocene palimpsests: an interdisciplinary approach to the Châtelperronian and carnivore occupations at Cassenade (Dordogne, France). Paleoanthropology 2019, 362–388. https://doi.org/10.4207/PA.2019.ART136

Gilead, I. (2002). Too many notes? Virtual recording of artifacts provenance. In: Niccolucci, F. (Ed.). Virtual Archaeology: Proceedings of the VAST Euroconference, Arezzo 24–25 November 2000. BAR International Series 1075, Archaeopress, Oxford, pp. 41–44.

Royer, A., Discamps, E., Plutniak, S. and Thomas, M. (2023). SEAHORS: Spatial Exploration of ArcHaeological Objects in R Shiny Zenodo, 7957154, ver. 2 peer-reviewed and recommended by Peer Community in Archaeology. https://doi.org/10.5281/zenodo.7929462

Yeshurun, R. (2022). Intra-site analysis of repeatedly occupied camps: Sacrificing “resolution” to get the story. In: Clark A.E., Gingerich J.A.M. (Eds.). Intrasite Spatial Analysis of Mobile and Semisedentary Peoples: Analytical Approaches to Reconstructing Occupation History. University of Utah Press, pp. 27–35. 

 

SEAHORS: Spatial Exploration of ArcHaeological Objects in R ShinyROYER, Aurélien, DISCAMPS, Emmanuel, PLUTNIAK, Sébastien, THOMAS, Marc<p style="text-align: justify;">This paper presents SEAHORS, an R shiny application available as an R package, dedicated to the intra-site spatial analysis of piece-plotted archaeological remains. This open-source script generates 2D and 3D scatte...Computational archaeology, Spatial analysis, Theoretical archaeologyReuven Yeshurun2023-02-24 16:01:44 View
03 Feb 2024
article picture

Digital surface models of crops used in archaeological feature detection – a case study of Late Neolithic site Tomašanci-Dubrava in Eastern Croatia

What lies on top lies also beneath? Connecting crop surface modelling to buried archaeology mapping.

Recommended by ORCID_LOGO based on reviews by Ian Moffat and Geert Verhoeven

This paper (Sosic et al. 2024) explores the Neolithic landscape of the Sopot culture in Đakovština, Eastern Slavonija, revealing a network of settlements through a multi-faceted approach that combines aerial archaeology, magnetometry, excavation, and field survey. This strategy facilitates scalable research tailored to the particularities of each site and allows for improved representations of buried archaeology with minimal intrusion. 

Using the site of Tomašanci-Dubrava as an example of the overall approach, the study further explores the use of drone imagery for 3D surface modeling, revealing a consistent correlation between crop surface elevation during full plant growth and ground terrain after ploughing, attributed to subsurface archaeological features. Results are correlated with magnetic survey and test-pitting data to validate the micro-topography and clarify the relationship between different subsurface structures.

The results obtained are presented in a comprehensive way, including their source data, and are contextualized in relation to conventional cropmark detection approaches and expectations. I found this aspect very interesting, since the crop surface and terrain models contradict typical or textbook examples of cropmark detection, where the vegetation is projected to appear higher in ditches and lower in areas with buried archaeology (Renfrew & Bahn 2016, 82). Regardless, the findings suggest the potential for broader applications of crop surface or canopy height modelling in landscape wide surveys, utilizing ALS data or aerial photographs.  

It seems then that the authors make a valid argument for a layered approach in landscape-based site detection, where aerial imagery can be used to accurately map the topography of areas of interest, which can then be further examined at site scale using more demanding methods, such as geophysical survey and excavation. This scalability enhances the research's relevance in broader archaeological and geographical contexts and renders it a useful example in site detection and landscape-scale mapping.

References

Renfrew, C. and Bahn, P. (2016). Archaeology: theories, methods and practice. Thames and Hudson. 

Sosic Klindzic, R., Vuković, M., Kalafatić, H. and Šiljeg, B. (2024). Digital surface models of crops used in archaeological feature detection – a case study of Late Neolithic site Tomašanci-Dubrava in Eastern Croatia, Zenodo, 7970703, ver. 4 peer-reviewed and recommended by Peer Community in Archaeology. https://doi.org/10.5281/zenodo.7970703

Digital surface models of crops used in archaeological feature detection – a case study of Late Neolithic site Tomašanci-Dubrava in Eastern CroatiaSosic Klindzic Rajna; Vuković Miroslav; Kalafatić Hrvoje; Šiljeg Bartul<p>This paper presents the results of a study on the neolithic landscape of the Sopot culture in the area of Đakovština in Eastern Slavonija. A vast network of settlements was uncovered using aerial archaeology, which was further confirmed and chr...Landscape archaeology, Neolithic, Remote sensing, Spatial analysisMarkos Katsianis2023-09-01 12:57:04 View