7
views
0
recommends
+1 Recommend
0 collections
    0
    shares
      • Record: found
      • Abstract: found
      • Article: found
      Is Open Access

      On the impact of Citizen Science-derived data quality on deep learning based classification in marine images

      research-article

      Read this article at

      Bookmark
          There is no author summary for this article yet. Authors can add summaries to their articles on ScienceOpen to make them more accessible to a non-specialist audience.

          Abstract

          The evaluation of large amounts of digital image data is of growing importance for biology, including for the exploration and monitoring of marine habitats. However, only a tiny percentage of the image data collected is evaluated by marine biologists who manually interpret and annotate the image contents, which can be slow and laborious. In order to overcome the bottleneck in image annotation, two strategies are increasingly proposed: “citizen science” and “machine learning”. In this study, we investigated how the combination of citizen science, to detect objects, and machine learning, to classify megafauna, could be used to automate annotation of underwater images. For this purpose, multiple large data sets of citizen science annotations with different degrees of common errors and inaccuracies observed in citizen science data were simulated by modifying “gold standard” annotations done by an experienced marine biologist. The parameters of the simulation were determined on the basis of two citizen science experiments. It allowed us to analyze the relationship between the outcome of a citizen science study and the quality of the classifications of a deep learning megafauna classifier. The results show great potential for combining citizen science with machine learning, provided that the participants are informed precisely about the annotation protocol. Inaccuracies in the position of the annotation had the most substantial influence on the classification accuracy, whereas the size of the marking and false positive detections had a smaller influence.

          Related collections

          Most cited references10

          • Record: found
          • Abstract: not found
          • Article: not found

          A survey of deep neural network architectures and their applications

            Bookmark
            • Record: found
            • Abstract: found
            • Article: found
            Is Open Access

            What Is Citizen Science? – A Scientometric Meta-Analysis

            Context The concept of citizen science (CS) is currently referred to by many actors inside and outside science and research. Several descriptions of this purportedly new approach of science are often heard in connection with large datasets and the possibilities of mobilizing crowds outside science to assists with observations and classifications. However, other accounts refer to CS as a way of democratizing science, aiding concerned communities in creating data to influence policy and as a way of promoting political decision processes involving environment and health. Objective In this study we analyse two datasets (N = 1935, N = 633) retrieved from the Web of Science (WoS) with the aim of giving a scientometric description of what the concept of CS entails. We account for its development over time, and what strands of research that has adopted CS and give an assessment of what scientific output has been achieved in CS-related projects. To attain this, scientometric methods have been combined with qualitative approaches to render more precise search terms. Results Results indicate that there are three main focal points of CS. The largest is composed of research on biology, conservation and ecology, and utilizes CS mainly as a methodology of collecting and classifying data. A second strand of research has emerged through geographic information research, where citizens participate in the collection of geographic data. Thirdly, there is a line of research relating to the social sciences and epidemiology, which studies and facilitates public participation in relation to environmental issues and health. In terms of scientific output, the largest body of articles are to be found in biology and conservation research. In absolute numbers, the amount of publications generated by CS is low (N = 1935), but over the past decade a new and very productive line of CS based on digital platforms has emerged for the collection and classification of data.
              Bookmark
              • Record: found
              • Abstract: found
              • Article: not found

              From principles to practice: a spatial approach to systematic conservation planning in the deep sea.

              Increases in the demand and price for industrial metals, combined with advances in technological capabilities have now made deep-sea mining more feasible and economically viable. In order to balance economic interests with the conservation of abyssal plain ecosystems, it is becoming increasingly important to develop a systematic approach to spatial management and zoning of the deep sea. Here, we describe an expert-driven systematic conservation planning process applied to inform science-based recommendations to the International Seabed Authority for a system of deep-sea marine protected areas (MPAs) to safeguard biodiversity and ecosystem function in an abyssal Pacific region targeted for nodule mining (e.g. the Clarion-Clipperton fracture zone, CCZ). Our use of geospatial analysis and expert opinion in forming the recommendations allowed us to stratify the proposed network by biophysical gradients, maximize the number of biologically unique seamounts within each subregion, and minimize socioeconomic impacts. The resulting proposal for an MPA network (nine replicate 400 × 400 km MPAs) covers 24% (1 440 000 km(2)) of the total CCZ planning region and serves as example of swift and pre-emptive conservation planning across an unprecedented area in the deep sea. As pressure from resource extraction increases in the future, the scientific guiding principles outlined in this research can serve as a basis for collaborative international approaches to ocean management.
                Bookmark

                Author and article information

                Contributors
                Role: ConceptualizationRole: Data curationRole: Formal analysisRole: InvestigationRole: MethodologyRole: SoftwareRole: ValidationRole: VisualizationRole: Writing – original draftRole: Writing – review & editing
                Role: ConceptualizationRole: Data curationRole: Formal analysisRole: InvestigationRole: MethodologyRole: ValidationRole: Writing – review & editing
                Role: ConceptualizationRole: Formal analysisRole: InvestigationRole: MethodologyRole: ValidationRole: Writing – review & editing
                Role: ConceptualizationRole: Formal analysisRole: Funding acquisitionRole: InvestigationRole: Project administrationRole: ResourcesRole: SupervisionRole: ValidationRole: Writing – review & editing
                Role: ConceptualizationRole: Formal analysisRole: Funding acquisitionRole: InvestigationRole: Project administrationRole: ResourcesRole: SupervisionRole: ValidationRole: VisualizationRole: Writing – original draftRole: Writing – review & editing
                Role: Editor
                Journal
                PLoS One
                PLoS ONE
                plos
                plosone
                PLoS ONE
                Public Library of Science (San Francisco, CA USA )
                1932-6203
                2019
                12 June 2019
                : 14
                : 6
                : e0218086
                Affiliations
                [1 ] Biodata Mining Group, Faculty of Technology, Bielefeld University, Bielefeld, Germany
                [2 ] National Oceanography Centre, University of Southampton Waterfront Campus, Southampton, United Kingdom
                New York University School of Medicine, UNITED STATES
                Author notes

                Competing Interests: The GPU donation from NVIDIA Corporation does not introduce a competing interest. This does not alter our adherence to PLOS ONE policies on sharing data and materials.

                Author information
                http://orcid.org/0000-0003-1857-5040
                http://orcid.org/0000-0001-9667-2917
                http://orcid.org/0000-0001-5218-1649
                Article
                PONE-D-18-30321
                10.1371/journal.pone.0218086
                6561570
                31188894
                3b2d4c9b-01b6-4373-b482-c956329d8e24
                © 2019 Langenkämper et al

                This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

                History
                : 19 October 2018
                : 25 May 2019
                Page count
                Figures: 8, Tables: 1, Pages: 16
                Funding
                We thank NVIDIA Corporation ( www.nvidia.com) for donating the GPU used in this project to TWN. The funder provided no support in the form of salaries for any author and did not have any additional role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript. DL has received funding by Projektträger Jülich (grant no 03F0707C), as well as ESL, DOBJ under the framework of JPI Oceans. ESL, DOBJ, BH have received funding from the European Union Seventh Framework Programme (FP7/2007-2013) under the MIDAS (Managing Impacts of DeepseA reSource exploitation) project, grant agreement 603418. Funding was also provided from the UK Natural Environment Research Council through National Capability funding to NOC. The funder provided support in the form of salaries for authors DL, ESL, DOBJ, BH, but did not have any additional role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript. The specific roles of these authors are articulated in the ‘author contributions’ section”.
                Categories
                Research Article
                Computer and Information Sciences
                Artificial Intelligence
                Machine Learning
                Deep Learning
                Science Policy
                Science and Technology Workforce
                Careers in Research
                Scientists
                People and Places
                Population Groupings
                Professions
                Scientists
                Science Policy
                Science and Technology Workforce
                Citizen Science
                Biology and Life Sciences
                Marine Biology
                Earth Sciences
                Marine and Aquatic Sciences
                Marine Biology
                Computer and Information Sciences
                Artificial Intelligence
                Machine Learning
                Earth Sciences
                Marine and Aquatic Sciences
                Research and Analysis Methods
                Imaging Techniques
                Biology and Life Sciences
                Organisms
                Eukaryota
                Animals
                Vertebrates
                Fish
                Marine Fish
                Biology and Life Sciences
                Marine Biology
                Marine Fish
                Earth Sciences
                Marine and Aquatic Sciences
                Marine Biology
                Marine Fish
                Custom metadata
                The data can be accessed at https://doi.org/10.5281/zenodo.3236009.

                Uncategorized
                Uncategorized

                Comments

                Comment on this article