• Login
     
    View Item 
    •   ORFEO Home
    • Royal Museum for Central Africa
    • RMCA publications
    • View Item
    •   ORFEO Home
    • Royal Museum for Central Africa
    • RMCA publications
    • View Item
    JavaScript is disabled for your browser. Some features of this site may not work without it.

    Fully convolutional networks for land cover classification from historical panchromatic aerial photographs

    Authors
    Mboga, N.
    Grippa, T.
    Georganos, S.
    Vanhuysse, S.
    Smets, B.
    Dewitte, O.
    Wolff, E.
    Lennert, M.
    Show allShow less
    Discipline
    Earth and related Environmental sciences
    Subject
    Natural hazards
    Audience
    Scientific
    Date
    2020
    Metadata
    Show full item record
    Description
    Historical aerial photographs provide salient information on the historical state of the landscape. The exploitation of these archives is often limited by accessibility and the time-consuming process of digitizing the analogue copies at a high resolution and processing them with a proper photogrammetric workflow. Furthermore, these data are characterised by limited spectral information since it occurs very often in a single band. Our work presents a first application of deep learning for the extraction of land cover from historical aerial panchromatic photographs of the African cities of Goma, Bukavu and Bujumbura. We evaluate the suitability of deep learning for land cover generation from a challenging dataset of photographs from the 1940s and 1950s that covers large geographical extents and is characterised by radiometric variations between dates and locations. A fully convolutional approach is investigated by considering two network architectures with different strategies of exploiting contextual information: one used atrous convolutional layers without downsampling, whereas the second network has both downsampling and learned upsampling convolutional layers (U-NET). The networks are trained to detect three main classes namely, buildings, high vegetation and a mixed class of bare land and low vegetation class. High overall accuracies of >90% in Goma-Gisenyi and Bukavu, and >85% in Bujumbura are obtained. This work provides a novel methodology that outperforms a baseline standard machine learning classifier for the exploitation of the vast archives of historical aerial photographs that can aid long-term environmental baseline studies. Future work will entail developing domain adaptation strategies in order to make the trained network robust for different image mosaics.
    Citation
    Mboga, N.; Grippa, T.; Georganos, S.; Vanhuysse, S.; Smets, B.; Dewitte, O.; Wolff, E.; Lennert, M. (2020). Fully convolutional networks for land cover classification from historical panchromatic aerial photographs. , ISPRS Journal of Photogrammetry and Remote Sensing, Vol. 167, 385-395, DOI: 10.1016/j.isprsjprs.2020.07.005.
    Identifiers
    uri: https://orfeo.belnet.be/handle/internal/12529
    doi: http://dx.doi.org/10.1016/j.isprsjprs.2020.07.005
    url: https://www.sciencedirect.com/science/article/abs/pii/S0924271620301921
    Type
    Article
    Peer-Review
    Yes
    Language
    eng
    Links
    NewsHelpdeskBELSPO OA Policy

    Browse

    All of ORFEOCommunities & CollectionsBy Issue DateAuthorsTitlesDisciplinesThis CollectionBy Issue DateAuthorsTitlesDisciplines
     

    DSpace software copyright © 2002-2016  DuraSpace
    Send Feedback | Cookie Information
    Theme by 
    Atmire NV