Faces as Lighting Probes via Unsupervised Deep Highlight Extraction

  title={Faces as Lighting Probes via Unsupervised Deep Highlight Extraction},
  author={Renjiao Yi and Chenyang Zhu and Ping Tan and Stephen Lin},
  • Renjiao Yi, Chenyang Zhu, +1 author Stephen Lin
  • Published in ECCV 2018
  • Computer Science
  • We present a method for estimating detailed scene illumination using human faces in a single image. [...] Key Method Based on the observation that faces can exhibit strong highlight reflections from a broad range of lighting directions, we propose a deep neural network for extracting highlights from faces, and then trace these reflections back to the scene to acquire the environment map.Expand Abstract
    CoSegNet: Deep Co-Segmentation of 3D Shapes with Group Consistency Loss
    • 6
    AdaCoSeg: Adaptive Shape Co-Segmentation With Group Consistency Loss
    • 2
    • PDF
    Leveraging Multi-view Image Sets for Unsupervised Intrinsic Image Decomposition and Highlight Separation
    Object-based Illumination Estimation with Rendering-aware Neural Networks
    IllumiNet: Transferring Illumination from Planar Surfaces to Virtual Objects in Augmented Reality
    Learning Illumination from Diverse Portraits


    Publications referenced by this paper.
    Image quality assessment: from error visibility to structural similarity
    • 23,663
    • Highly Influential
    • PDF
    Face detection, pose estimation, and landmark localization in the wild
    • 1,933
    • Highly Influential
    • PDF
    Using color to separate reflection components
    • 1,276
    Illumination for computer generated pictures
    • 3,108
    • PDF
    Direct Sparse Odometry
    • 953
    • PDF
    A 3D Face Model for Pose and Illumination Invariant Face Recognition
    • 702
    • Highly Influential
    • PDF
    Shape, Illumination, and Reflectance from Shading
    • 428
    • PDF
    A signal-processing framework for inverse rendering
    • 564
    • Highly Influential
    • PDF
    An iterative technique for the rectification of observed distributions
    • 2,868
    • PDF