Unsupervised Salient Object Detection with Spectral Cluster Voting (2022-03-23T00:00:00.000000Z)

TL;DR

This paper proposes a simple but effective winner-takes-all voting mechanism for selecting the salient masks, leveraging object priors based on framing and distinctiveness and trains a salient object detector, termed SELF-MASK, which outperforms prior approaches on three unsupervised SOD benchmarks.

Abstract

In this paper, we tackle the challenging task of unsupervised salient object detection (SOD) by leveraging spectral clustering on self-supervised features. We make the following contributions: (i) We revisit spectral clustering and demonstrate its potential to group the pixels of salient objects across various self-supervised features, e.g., Mo-Cov2, SwAV, and DINO; (ii) Given mask proposals from multiple applications of spectral clustering on image features computed from different self-supervised models, we propose a simple but effective winner-takes-all voting mechanism for selecting the salient masks, leveraging object priors based on framing and distinctiveness; (iii) Using the selected object segmentation as pseudo groundtruth masks, we train a salient object detector, termed SELF-MASK, which outperforms prior approaches on three un-supervised SOD benchmarks. Code is publicly available at https://github.com/NoelShin/selfmask.

Authors

Weidi Xie

19 papers

Samuel Albanie

18 papers

Gyungin Shin

2 papers

TL;DR

Abstract

Authors

References76 items

Self-Supervised Transformers for Unsupervised Object Discovery using Normalized Cut

Localizing Objects with Self-Supervised Transformers and no Labels

Per-Pixel Classification is Not All You Need for Semantic Segmentation

BEiT: BERT Pre-Training of Image Transformers

Intriguing Properties of Vision Transformers

Finding an Unsupervised Image Segmenter in Each of Your Deep Generative Models

Emerging Properties in Self-Supervised Vision Transformers

An Empirical Study of Training Self-Supervised Vision Transformers

Tokens-to-Token ViT: Training Vision Transformers from Scratch on ImageNet

Training data-efficient image transformers & distillation through attention

An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale

DEAL: Difficulty-aware Active Learning for Semantic Segmentation

Unsupervised Learning of Visual Features by Contrasting Cluster Assignments

Object Segmentation Without Labels with Large-Scale Generative Models

End-to-End Object Detection with Transformers

Semi-Supervised Semantic Segmentation With Cross-Consistency Training

Unsupervised Discovery of Interpretable Directions in the GAN Latent Space

Momentum Contrast for Unsupervised Visual Representation Learning

Self-labelling via simultaneous clustering and representation learning

EGNet: Edge Guidance Network for Salient Object Detection

Unsupervised Object Segmentation by Redrawing

Object Discovery with a Copy-Pasting GAN

Emergence of Object Segmentation in Perturbed Generative Models

A Simple Pooling-Based Design for Real-Time Salient Object Detection

Salient Object Detection in the Deep Learning Era: An In-Depth Survey

Cross Pixel Optical Flow Similarity for Self-Supervised Learning

Deep Clustering for Unsupervised Learning of Visual Features

Unsupervised Feature Learning via Non-parametric Instance Discrimination

Deep Unsupervised Saliency Detection: A Multiple Noisy Labeling Perspective

Encoder-Decoder with Atrous Separable Convolution for Semantic Image Segmentation

Mix-and-Match Tuning for Self-Supervised Semantic Segmentation

Decoupled Weight Decay Regularization

Supervision by Fusion: Towards Unsupervised Learning of Deep Salient Object Detector

Focal Loss for Dense Object Detection

Learning to Detect Salient Objects with Image-Level Supervision

Attention is All you Need

Colorization as a Proxy Task for Visual Understanding

V-Net: Fully Convolutional Neural Networks for Volumetric Medical Image Segmentation

Exploiting Surroundedness for Saliency Detection: A Boolean Map Approach

Hierarchical Image Saliency Detection on Extended CSSD

Unsupervised Learning of Visual Representations by Solving Jigsaw Puzzles

Deep Residual Learning for Image Recognition

Minimum Barrier Salient Object Detection at 80 FPS

Learning visual groups from co-occurrences in space and time

The Fast Bilateral Solver

A weighted sparse coding framework for saliency detection

Unsupervised Visual Representation Learning by Context Prediction

Visual saliency based on multiscale deep features

Saliency Optimization from Robust Background Detection

Multiscale Combinatorial Grouping

Saliency Detection via Graph-Based Manifold Ranking

Hierarchical Saliency Detection

Geodesic Saliency Using Background Priors

Saliency filters: Contrast based filtering for salient region detection

Center-surround divergence of feature statistics for salient object detection

Scale and object aware image retargeting for thumbnail browsing

Global contrast based salient region detection

Importance filtering for image retargeting

Design and perceptual validation of performance measures for salient object segmentation

SUN database: Large-scale scene recognition from abbey to zoo

Context-aware saliency detection

Learning to predict where humans look

Frequency-tuned salient region detection

ImageNet: A large-scale hierarchical image database

A tutorial on spectral clustering

Learning to Detect A Salient Object

Automatic thumbnail cropping and its effectiveness

A Model of Saliency-Based Visual Attention for Rapid Scene Analysis

Normalized cuts and image segmentation

Support-Vector Networks

Least squares quantization in PCM

DeepUSPS: Deep Robust Unsupervised Saliency Prediction via Self-supervision

GENERATIVE ADVERSARIAL NETS

A framework for visual saliency detection with applications to image thumbnailing

To appear in the ACM SIGGRAPH conference proceedings Improved Seam Carving for Video Retargeting

Ieee Transactions on Pattern Analysis and Machine Intelligence 1 Discriminative Unsupervised Feature Learning with Exemplar Convolutional Neural Networks