Toward a Taxonomy and Computational Models of Abnormalities in Images

@article{Saleh2016TowardAT,
  title={Toward a Taxonomy and Computational Models of Abnormalities in Images},
  author={B. Saleh and A. Elgammal and J. Feldman and A. Farhadi},
  journal={ArXiv},
  year={2016},
  volume={abs/1512.01325}
}
  • B. Saleh, A. Elgammal, +1 author A. Farhadi
  • Published 2016
  • Computer Science, Mathematics
  • ArXiv
  • The human visual system can spot an abnormal image, and reason about what makes it strange. [...] Key Method We design human subject experiments to discover a coarse taxonomy of the reasons for abnormality. Our experiments reveal three major categories of abnormality: object-centric, scene-centric, and contextual. Based on this taxonomy, we propose a comprehensive computational model that can predict all different types of abnormality in images and outperform prior arts in abnormality recognition.Expand Abstract

    Figures, Tables, and Topics from this paper.

    Automatic Understanding of Image and Video Advertisements
    • 53
    • Open Access
    Sample-efficient image segmentation through recurrence
    • 6
    Multi-view manifold learning with locality alignment
    • 32
    Landmark detection with surprise saliency using convolutional neural networks
    How Convolutional Neural Networks Remember Art
    • 1
    • Open Access

    References

    Publications referenced by this paper.
    SHOWING 1-10 OF 47 REFERENCES
    Very Deep Convolutional Networks for Large-Scale Image Recognition
    • 39,028
    • Open Access
    Recognition-by-components: a theory of human image understanding.
    • 5,378
    • Open Access
    Rich Feature Hierarchies for Accurate Object Detection and Semantic Segmentation
    • 11,341
    • Open Access
    SUN database: Large-scale scene recognition from abbey to zoo
    • 1,973
    • Open Access
    OverFeat: Integrated Recognition, Localization and Detection using Convolutional Networks
    • 3,605
    • Open Access
    Learning Deep Features for Scene Recognition using Places Database
    • 2,162
    • Open Access
    ImageNet classification with deep convolutional neural networks
    • 51,807
    • Open Access
    ImageNet Classification with Deep Convolutional Neural Networks
    • 51,807
    • Open Access
    Basic objects in natural categories
    • 4,623
    • Open Access
    The Role of Context for Object Detection and Semantic Segmentation in the Wild
    • 619
    • Open Access