A framework for the development of a global standardised marine taxon reference image database (SMarTaR-ID) to support image-based analyses

Abstract

75Video and image data are regularly used in the field of benthic ecology to document 76 biodiversity. However, their use is subject to a number of challenges, principally the 77 identification of taxa within the images without associated physical specimens. The 78 challenge of applying traditional taxonomic keys to the identification of fauna from 79 images has led to the development of personal, group, or institution level reference 80 image catalogues of operational taxonomic units (OTUs) or morphospecies. Lack of 81 standardisation among these reference catalogues has led to problems with 82 observer bias and the inability to combine datasets across studies. In addition, lack 83 of a common reference standard is stifling efforts in the application of artificial 84 intelligence to taxon identification. Using the North Atlantic deep sea as a case 85 study, we propose a database structure to facilitate standardisation of 86 morphospecies image catalogues between research groups and support future use 87 in multiple front-end applications. We also propose a framework for coordination of 88 international efforts to develop reference guides for the identification of marine 89 species from images. The proposed structure follows the Darwin Core standard to 90 allow integration with existing databases. We suggest a management framework 91 where high-level taxonomic groups are curated by a regional team, consisting of 92 both end users and taxonomic experts. We identify a mechanism by which overall 93 quality of data within a common reference guide could be raised over the next 94 decade. Finally, we discuss the role of a common reference standard in advancing 95 marine ecology and supporting sustainable use of this ecosystem. 96 6 97 100 the sunlit Mediterranean seabed, for the first clear images to be produced [2]. 101 Following this, the use of underwater photography became widespread in shallow 102 seas, opening up this environment to a wider public (e.g. [3]). The first deep-sea 103 photograph was taken from the porthole of a bathysphere in the early 1930s [4] and 104 shortly after, the first self-contained deep-sea photographic systems were developed 105in the 1940s at the Woods Hole Oceanographic Institution [5,6]. Whilst there were 106 many good deep-sea photographs available between this time and the early 1970s 107 [7,8], few biologists studied them, as often no corresponding samples of animals 108 were taken, making identification difficult [9]. The notable exceptions to this [9,10, 11, 109 12, 13, 14] paved the way for photography to become established as an important 110 tool for the study of deep-water environments [15, 16, 17, 18, 19]. Today, with the 111 routine use of seafloor cameras, towed camera platforms, remotely operated and 112 autonomous underwater vehicles (ROVs and AUVs), photographic assessment of 113 marine fauna and faunal assemblages is a vital tool for research used by both 114 scientists and industry [20, 21, 22]. 115 Imaging is an important non-destructive tool for studying ...

Cold‐water coral assemblages on vertical walls from the Northeast Atlantic

“…Use of morphospecies enables the differentiation of taxa below the lowest taxonomic level to which an organism can be identified based on imagery alone, using features such as colour, growth form, branching pattern, ecological information (e.g., depth), etc. (Howell et al, ). As this approach complicates comparison between research groups or reuse of data, a reference image for each morphospecies (as well as the species matrix) is provided as Supporting Information.…”

A benthic substrate classification method for seabed images using deep learning: Application to management of deep‐sea coral reefs

“…Quantitative image data therefore provide an effective and nonextractive way of measuring properties of marine ecosystems (physical habitats and their biological inhabitants) including changes in status through time stemming from anthropogenic impacts and responses to management interventions. The use of image-derived data in marine environmental science has accelerated due to increased access to enabling technology and its greater affordability, and as international standards for classifying fauna are developed and adopted (Althaus et al, 2015;Howell et al, 2019).…”

“…Surveys using cameras provide a prospective means of improving knowledge of VME distributions because image‐based surveys offer valuable insight into fauna and habitats for rarely sampled deep‐sea ecosystem and can be used to inform spatial models that predict their distributions over broad areas (Baco et al, 2020; Buhl‐Mortensen et al, 2019). Image‐derived data have several attractive qualities including being nonextractive, quantitative, spatially scalable and universally effective for stakeholder engagement (Durden et al, 2016; Howell et al, 2019). Acquiring large volumes of high‐quality raw imagery is tractable (Untiedt et al, 2021), but the time‐consuming and labour‐intensive process of quantitatively annotating images to identify, consistently classify, and record even relatively simple features of interest such as substrates or VMEs creates a bottleneck that severely limits the volume of analysed data (Perkins et al, 2022).…”

High-Resolution Vertical Habitat Mapping of a Deep-Sea Cliff Offshore Greenland

“…Annotation of marine megafauna and estimation of the seabed area within each image were carried out using the online BIIGLE 2.0 platform (Langenkämper et al, 2017). A morphospecies approach was used to characterize the diversity of the epibenthic megafauna, as standardized taxonomic classification of species from deep-sea imagery is not always accurate (Howell et al, 2019).…”