Built-in foreground/background prior for weakly-supervised semantic segmentation

Fatemehsadat Saleh*, Mohammad Sadegh Ali Akbarian, Mathieu Salzmann, Lars Petersson, Stephen Gould, Jose M. Alvarez

*Corresponding author for this work

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

    102 Citations (Scopus)

    Abstract

    Pixel-level annotations are expensive and time consuming to obtain. Hence, weak supervision using only image tags could have a significant impact in semantic segmentation. Recently, CNN-based methods have proposed to fine-tune pre-trained networks using image tags. Without additional information, this leads to poor localization accuracy. This problem, however, was alleviated by making use of objectness priors to generate foreground/background masks. Unfortunately these priors either require training pixel-level annotations/bounding boxes, or still yield inaccurate object boundaries. Here, we propose a novel method to extract markedly more accurate masks from the pre-trained network itself, forgoing external objectness modules. This is accomplished using the activations of the higher-level convolutional layers, smoothed by a dense CRF. We demonstrate that our method, based on these masks and a weakly-supervised loss, outperforms the state-of-the-art tag-based weakly-supervised semantic segmentation techniques. Furthermore, we introduce a new form of inexpensive weak supervision yielding an additional accuracy boost.

    Original languageEnglish
    Title of host publicationComputer Vision - 14th European Conference, ECCV 2016, Proceedings
    EditorsBastian Leibe, Jiri Matas, Nicu Sebe, Max Welling
    PublisherSpringer Verlag
    Pages413-432
    Number of pages20
    ISBN (Print)9783319464831
    DOIs
    Publication statusPublished - 2016
    Event14th European Conference on Computer Vision, ECCV 2016 - Amsterdam, Netherlands
    Duration: 8 Oct 201616 Oct 2016

    Publication series

    NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
    Volume9912 LNCS
    ISSN (Print)0302-9743
    ISSN (Electronic)1611-3349

    Conference

    Conference14th European Conference on Computer Vision, ECCV 2016
    Country/TerritoryNetherlands
    CityAmsterdam
    Period8/10/1616/10/16

    Fingerprint

    Dive into the research topics of 'Built-in foreground/background prior for weakly-supervised semantic segmentation'. Together they form a unique fingerprint.

    Cite this