Self-Supervised Transformers for Unsupervised Object Discovery using Normalized Cut

Transformers trained with self-supervised learning using self-distillation loss (DINO) have been shown to produce attention maps that highlight salient foreground objects. In this paper, we demonstrate a graph-based approach that uses the self-supervised transformer features to discover an object from an image. Visual tokens are viewed as nodes in a weighted graph with edges representing a connectivity score based on the similarity of tokens. Foreground objects can then be segmented using a normalized graph-cut to group self-similar regions. We solve the graph-cut problem using spectral clustering with generalized eigen-decomposition and show that the second smallest eigenvector provides a cutting solution since its absolute value indicates the likelihood that a token belongs to a foreground object. Despite its simplicity, this approach significantly boosts the performance of unsupervised object discovery: we improve over the recent state of the art LOST by a margin of 6.9%, 8.1%, and 8.1% respectively on the VOC07, VOC12, and COCO20K. The performance can be further improved by adding a second stage class-agnostic detector (CAD). Our proposed method can be easily extended to unsupervised saliency detection and weakly supervised object detection. For unsupervised saliency detection, we improve IoU for 4.9%, 5.2%, 12.9% on ECSSD, DUTS, DUT-OMRON respectively compared to previous state of the art. For weakly supervised object detection, we achieve competitive performance on CUB and ImageNet.

PDF Abstract CVPR 2022 PDF CVPR 2022 Abstract
Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Single-object discovery COCO_20k TokenCut CorLoc 58.8 # 5
Single-object discovery COCO_20k TokenCut + CAD CorLoc 62.6 # 4
Weakly-Supervised Object Localization CUB TokenCut Top-1 Localization Accuracy 72.9 # 1
Weakly-Supervised Object Localization CUB-200-2011 TokenCut Top-1 Localization Accuracy 72.9 # 3
Unsupervised Saliency Detection DUT-OMRON TokenCut Accuracy 89.7 # 3
IoU 61.8 # 3
maximal F-measure 69.7 # 3
Unsupervised Saliency Detection DUTS TokenCut Accuracy 91.4 # 3
IoU 62.4 # 3
maximal F-measure 75.5 # 3
Unsupervised Saliency Detection ECSSD TokenCut Accuracy 93.4 # 3
IoU 77.2 # 3
maximal F-measure 87.4 # 3
Weakly-Supervised Object Localization ImageNet TokenCut GT-known localization accuracy 65.4 # 4
Top-1 Localization Accuracy 52.3 # 4