Superpixel Sampling Networks

@inproceedings{Jampani2018SuperpixelSN,
  title={Superpixel Sampling Networks},
  author={V. Jampani and Deqing Sun and Ming-Yu Liu and Ming-Hsuan Yang and Jan Kautz},
  booktitle={ECCV},
  year={2018}
}
Superpixels provide an efficient low/mid-level representation of image data, which greatly reduces the number of image primitives for subsequent vision tasks. Existing superpixel algorithms are not differentiable, making them difficult to integrate into otherwise end-to-end trainable deep neural networks. We develop a new differentiable model for superpixel sampling that leverages deep networks for learning superpixel segmentation. The resulting Superpixel Sampling Network (SSN) is end-to-end… 
SIN: Superpixel Interpolation Network
TLDR
A deep learning-based superpixel segmentation algorithm SIN which can be integrated with downstream tasks in an end-to-end way and benefit the superpixel-based community is proposed.
Superpixel Segmentation With Fully Convolutional Networks
TLDR
A novel method that employs a simple fully convolutional network to predict superpixels on a regular image grid and develops a downsampling/upsampling scheme for deep networks with the goal of generating high-resolution outputs for dense prediction tasks.
Deep Superpixel Convolutional Network for Image Recognition
TLDR
This paper proposes an end-to-end trainable superpixel convolutional network, named SPNet, to learn high-level representation on image superpixel primitives, and treats irregular superpixel lattices as a 2D point cloud.
End-to-end trainable network for superpixel and image segmentation
Implicit Integration of Superpixel Segmentation into Fully Convolutional Networks
TLDR
This paper proposes a way to implicitly integrate a superpixel scheme into CNNs, which makes it easy to use superpixels with CNNs in an end-to-end fashion, and preserves detailed information such as object boundaries in the form ofsuperpixels even when the model contains downsampling layers.
HERS Superpixels: Deep Affinity Learning for Hierarchical Entropy Rate Segmentation
TLDR
This work proposes a two-stage graph-based framework for superpixel segmentation called Hierarchical Entropy Rate Segmentation (HERS), which builds a hierarchical tree structure that can produce any number of highly adaptive superpixels instantaneously.
Superpixel Segmentation Via Convolutional Neural Networks with Regularized Information Maximization
  • Teppei Suzuki
  • Computer Science, Mathematics
    ICASSP 2020 - 2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)
  • 2020
TLDR
This work proposes an unsupervised superpixel segmentation method by optimizing a randomly-initialized convolutional neural network (CNN) in inference time and verifies the advantages quantitatively and qualitatively on BSDS500 dataset.
Superpixel segmentation with attention convolution neural network
TLDR
In this method, a more complex convolutional neural network with an attention module is applied to superpixel segmentation, and the addition of the attention module allows for more accurate chunking of the images, thus yielding more comprehensive and detailed segmentation results.
Rethinking Unsupervised Neural Superpixel Segmentation
TLDR
This paper proposes three key elements to improve the efficacy of unsupervised learning for superpixel segmentation via CNNs: the similarity of the soft superpixelated image compared to the input image, the enhancement and consideration of object edges and boundaries and a modified architecture based on atrous convolution, which allow for a wider field of view.
Learning Based SLIC Superpixel Generation and Image Segmentation
TLDR
An advanced superpixel generation algorithm that combines the advantages of conventional methods and modern neural network techniques is proposed and it is found that the feature generated by neural networks also provide useful information for superpixel assignment.
...
...

References

SHOWING 1-10 OF 46 REFERENCES
Learning Superpixels with Segmentation-Aware Affinity Loss
TLDR
This work proposes a segmentation-aware affinity learning approach for superpixel segmentation with a new loss function that takes the segmentation error into account for affinity learning and develops the Pixel Affinity Net for affinity prediction.
SLIC Superpixels Compared to State-of-the-Art Superpixel Methods
TLDR
A new superpixel algorithm is introduced, simple linear iterative clustering (SLIC), which adapts a k-means clustering approach to efficiently generate superpixels and is faster and more memory efficient, improves segmentation performance, and is straightforward to extend to supervoxel generation.
SuperCNN: A Superpixelwise Convolutional Neural Network for Salient Object Detection
TLDR
A novel superpixelwise convolutional neural network approach, called SuperCNN, is proposed to learn the internal representations of saliency in an efficient manner, and can robustly detect salient objects and outperforms the state-of-the-art methods on three benchmark datasets.
Superpixels and Supervoxels in an Energy Optimization Framework
TLDR
This work forms the superpixel partitioning problem in an energy minimization framework, and explores variations of the basic energy, which allow a trade-off between a less regular tessellation but more accurate boundaries or better efficiency.
Superpixel Convolutional Networks Using Bilateral Inceptions
TLDR
A new “bilateral inception” module that can be inserted in existing CNN architectures and performs bilateral filtering, at multiple feature-scales, between superpixels in an image.
SEEDS: Superpixels Extracted Via Energy-Driven Sampling
TLDR
A robust and fast to evaluate energy function is defined, based on enforcing color similarity between the boundaries and the superpixel color histogram, which is able to achieve a performance comparable to the state-of-the-art, but in real-time on a single Intel i7 CPU at 2.8 GHz.
Superpixels: An evaluation of the state-of-the-art
TurboPixels: Fast Superpixels Using Geometric Flows
TLDR
A geometric-flow-based algorithm for computing a dense oversegmentation of an image, often referred to as superpixels, which yields less undersegmentation than algorithms that lack a compactness constraint while offering a significant speedup over N-cuts, which does enforce compactness.
Waterpixels
TLDR
This work proposes waterpixels as a general strategy for generating superpixels which relies on the marker controlled watershed transformation and introduces a spatially regularized gradient to achieve a tunable tradeoff between the superpixel regularity and the adherence to object boundaries.
Object detection by labeling superpixels
TLDR
This paper takes object detection as a multi-label superpixel labeling problem by minimizing an energy function and uses the data cost term to capture the appearance, smooth cost terms to encode the spatial context and label costterm to favor compact detection.
...
...