Deepwater Archaeological Survey: An Interdisciplinary and Complex Process

Abstract

This chapter introduces several state of the art techniques that could help to make deep underwater archaeological photogrammetric surveys easier, faster, more accurate, and to provide more visually appealing representations in 2D and 3D for both experts and public. We detail how the 3D captured data is analysed and then represented using ontologies, and how this facilitates interdisciplinary interpretation and cooperation. Towards more automation, we present a new method that adopts a deep learning approach for the detection and the recognition of objects of interest, amphorae for example. In order to provide more readable, direct and clearer illustrations, we describe several techniques that generate different styles of sketches out of orthophotos developed using neural networks. In the same direction, we present the Non-Photorealistic Rendering (NPR) technique, which converts a 3D model into a more readable 2D representation that is more useful to communicate and simplifies the identification of objects of interest. Regarding public dissemination, we demonstrate how recent advances in virtual reality to provide an accurate, high resolution, amusing and appropriate visualization tool that offers the public the possibility to 'visit' an unreachable archaeological site. Finally, we conclude by introducing the plenoptic approach, a new promising technology that can change the future of the photogrammetry by making it easier and less time consuming and that allows a user to create a 3D model using only one camera shot. Here, we introduce the concepts, the developing process, and some results, which we obtained with underwater imaging.

Archaeologic Machine Learning for Shipwreck Detection Using Lidar and Sonar

“…Nayak et al use side scan sonar collected by AUV, which limits the geographic extent of the project. Additionally, their model produces many false positives, as do the models introduced by Pasquet et al [12] and Drap et al [5]. The new model implementation presented here addresses false positives by integrating background topography in model training, achieving the highest precision value (90%) among these works by a large margin.…”

“…All other published related work focuses either on a very small study area that includes fewer than four shipwrecks [5,12,24] or used privately available side scan sonar [17,18,25,26]. While some of these projects produce shipwreck detection results comparable with the new work presented here, the work presented here differs in three significant ways: (1) the study area is very large, which enables rapid mapping of huge areas all at once; (2) the imagery is open source, which means this methodology could be replicated for additional study areas; (3) the imagery used includes a substantial amount of airborne lidar which means that imagery can be efficiently collected over a much larger area than with a shipborne or unmanned platform.…”

“…Their model was tested on two datasets: the model produced recall values between 80 and 97% and precision values between 29 and 33%. As is evidenced by the low precision values, the authors had problems with false positives similar to Pasquet et al [12] and Drap et al [5]. Xu et al [26] used a neural network to detect shipwrecks in sonar imagery with a very small training dataset.…”

“…Nonetheless, Shih et al's study shows that airborne bathymetric lidar is effective in shipwreck identification. Since 2017, Pasquet et al [12] and Drap et al [5] have been working together on a project to develop a deep learning approach for the detection and recognition of objects of interest using orthomosaics, with a focus on amphorae. Using transfer learning to help address their small training dataset size (transfer learning is the process of using as a model base layer the output of another model that has been pretrained on many thousands of images of other features), they test their approach to detect amphorae at a single shipwreck site located in 44 m of water.…”

“…Maritime shipwrecks represent a major resource to understand the human past over thousands of years, but underwater archaeology is expensive, misappropriated, and hazardous [1][2][3][4]. Therefore, a methodology to automatically map all potential shipwrecks over a large geographic area accurately and rapidly can help archaeologists prioritize site selection and plan excavation [5]. The model implementation presented here accurately detects shipwrecks in remotely sensed imagery collected from large areas of the coast of the continental United States, Alaska, and Puerto Rico.…”