Boundary Loss for Remote Sensing Imagery Semantic Segmentation

Alexey Bokhovkin, Evgeny Burnaev

    Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

    33 Citations (Scopus)

    Abstract

    In response to the growing importance of geospatial data, its analysis including semantic segmentation becomes an increasingly popular task in computer vision today. Convolutional neural networks are powerful visual models that yield hierarchies of features and practitioners widely use them to process remote sensing data. When performing remote sensing image segmentation, multiple instances of one class with precisely defined boundaries are often the case, and it is crucial to extract those boundaries accurately. The accuracy of segments boundaries delineation influences the quality of the whole segmented areas explicitly. However, widely-used segmentation loss functions such as BCE, IoU loss or Dice loss do not penalize misalignment of boundaries sufficiently. In this paper, we propose a novel loss function, namely a differentiable surrogate of a metric accounting accuracy of boundary detection. We can use the loss function with any neural network for binary segmentation. We performed validation of our loss function with various modifications of UNet on a synthetic dataset, as well as using real-world data (ISPRS Potsdam, INRIA AIL). Trained with the proposed loss function, models outperform baseline methods in terms of IoU score.

    Original languageEnglish
    Title of host publicationAdvances in Neural Networks – ISNN 2019 - 16th International Symposium on Neural Networks, ISNN 2019, Proceedings
    EditorsHuchuan Lu, Huajin Tang, Zhanshan Wang
    PublisherSpringer Verlag
    Pages388-401
    Number of pages14
    ISBN (Print)9783030228071
    DOIs
    Publication statusPublished - 2019
    Event16th International Symposium on Neural Networks, ISNN 2019 - Moscow, Russian Federation
    Duration: 10 Jul 201912 Jul 2019

    Publication series

    NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
    Volume11555 LNCS
    ISSN (Print)0302-9743
    ISSN (Electronic)1611-3349

    Conference

    Conference16th International Symposium on Neural Networks, ISNN 2019
    Country/TerritoryRussian Federation
    CityMoscow
    Period10/07/1912/07/19

    Keywords

    • Aerial imagery
    • Building detection
    • CNN
    • Computer vision
    • Deep learning
    • Loss function
    • Semantic segmentation

    Fingerprint

    Dive into the research topics of 'Boundary Loss for Remote Sensing Imagery Semantic Segmentation'. Together they form a unique fingerprint.

    Cite this