Linearized Multi-Sampling for Differentiable Image Transformation

@article{Jiang2019LinearizedMF,
  title={Linearized Multi-Sampling for Differentiable Image Transformation},
  author={Wei Jiang and Weiwei Sun and A. Tagliasacchi and Eduard Trulls and K. M. Yi},
  journal={2019 IEEE/CVF International Conference on Computer Vision (ICCV)},
  year={2019},
  pages={2988-2997}
}
  • Wei Jiang, Weiwei Sun, +2 authors K. M. Yi
  • Published 2019
  • Computer Science
  • 2019 IEEE/CVF International Conference on Computer Vision (ICCV)
  • We propose a novel image sampling method for differentiable image transformation in deep neural networks. The sampling schemes currently used in deep learning, such as Spatial Transformer Networks, rely on bilinear interpolation, which performs poorly under severe scale changes, and more importantly, results in poor gradient propagation. This is due to their strict reliance on direct neighbors. Instead, we propose to generate random auxiliary samples in the vicinity of each pixel in the sampled… CONTINUE READING
    Deep Image Spatial Transformation for Person Image Generation
    7
    P2Net: Patch-match and Plane-regularization for Unsupervised Indoor Depth Estimation
    MIST: Multiple Instance Spatial Transformer Networks
    SynSin: End-to-End View Synthesis From a Single Image
    14
    Trainable TV-L1 model as recurrent nets for low-level vision

    References

    Publications referenced by this paper.
    SHOWING 1-10 OF 45 REFERENCES
    Beyond Cartesian Representations for Local Descriptors
    15
    Learning to Match Aerial Images with Deep Attentive Architectures
    53
    LF-Net: Learning Local Features from Images
    110
    Deformable Convolutional Networks
    1101
    Hierarchical Spatial Transformer Network
    1
    U-Net: Convolutional Networks for Biomedical Image Segmentation
    14479
    Learning to Assign Orientations to Feature Points
    72
    Top-down Flow Transformer Networks
    1