SeGAN: Segmenting and Generating the Invisible

  title={SeGAN: Segmenting and Generating the Invisible},
  author={Kiana Ehsani and R. Mottaghi and Ali Farhadi},
  journal={2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  • Kiana Ehsani, R. Mottaghi, Ali Farhadi
  • Published 2018
  • Computer Science
  • 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition
  • Objects often occlude each other in scenes; Inferring their appearance beyond their visible parts plays an important role in scene understanding, depth estimation, object interaction and manipulation. In this paper, we study the challenging problem of completing the appearance of occluded objects. Doing so requires knowing which pixels to paint (segmenting the invisible parts of objects) and what color to paint them (generating the invisible parts). Our proposed novel solution, SeGAN, jointly… CONTINUE READING
    AI2-THOR: An Interactive 3D Environment for Visual AI
    • 189
    • Open Access
    Layer-structured 3D Scene Inference via View Synthesis
    • 46
    • Open Access
    Natural and Effective Obfuscation by Head Inpainting
    • 56
    • Open Access
    Image Segmentation Using Deep Learning: A Survey
    • 41
    • Open Access
    How Generative Adversarial Networks and Their Variants Work
    • 37
    • Open Access
    3D Scene Graph: A Structure for Unified Semantics, 3D Space, and Camera
    • 15
    • Open Access
    SAIL-VOS: Semantic Amodal Instance Level Video Object Segmentation – A Synthetic Dataset and Baselines
    • 7
    • Highly Influenced
    • Open Access
    How Generative Adversarial Nets and its variants Work: An Overview of GAN
    • 23
    Embodied Visual Recognition
    • 8
    • Open Access
    3D Point Cloud Generative Adversarial Network Based on Tree Structured Graph Convolutions
    • 21
    • Open Access


    Publications referenced by this paper.
    Microsoft COCO: Common Objects in Context
    • 10,251
    • Open Access
    Deep Residual Learning for Image Recognition
    • 50,145
    • Highly Influential
    • Open Access
    Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks
    • 18,378
    • Open Access
    Rich Feature Hierarchies for Accurate Object Detection and Semantic Segmentation
    • 11,412
    • Open Access
    You Only Look Once: Unified, Real-Time Object Detection
    • 8,730
    • Open Access
    U-Net: Convolutional Networks for Biomedical Image Segmentation
    • 15,428
    • Open Access
    Fully Convolutional Networks for Semantic Segmentation
    • 7,167
    • Open Access
    Image-to-Image Translation with Conditional Adversarial Networks
    • 6,250
    • Highly Influential
    • Open Access
    Fast R-CNN
    • 8,791
    • Open Access