Foreground Focus: Unsupervised Learning from Partially Matching Images

By Yong Jae Lee, Kristen Grauman. 2009

139 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Media Technology

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Media Technology

Citations in the corpus

Unsupervised Discovery of Mid-Level Discriminative Patches (2012, background): “…goals, our work is quite diﬀerent in that: 1) we do not explicitly aim to discover whole semantic units like objects or parts, 2) unlike [25,27,30], we do not assume a single object per image, 3) whereas in object discovery there is no separate training and test set, we explicitly aim to…”
Context as Supervisory Signal: Discovering Objects with Predictable Context (2014, background): “Proponents of unsupervised representation learning [1,2,3,4] and unsupervised object discovery [5,6,7,8,9,10,11,12] have long argued that these approaches have the potential to solve two fundamental problems with supervised methods.”
Multi-view Subspace Clustering (2015, methodology): “Just as [14], we select 7 classes, including tree, building, airplane, cow, face, car and bicycle.”
Weakly Supervised Localization and Learning with Generic Knowledge (2012, unknown): “…signiﬁcant progress in recent years, as several methods have tried to go beyond and experiment on more challenging datasets, such as ETHZ Shape Classes [Bagon et al, 2010, Lee and Grauman, 2009a], PASCAL VOC 06 [Chum and Zisserman, 2007, Kim and Torralba, 2009], and LabelMe [Russell et al, 2006].”
Parameter-Free Auto-Weighted Multiple Graph Learning: A Framework for Multiview Clustering and Semi-Supervised Classification (2016, methodology): “Following [Lee and Grauman, 2009], we select 7 classes composed of tree, building, airplane, cow, face,car, bicycle and each class has 30 images.”
Learning a hierarchy of discriminative space-time neighborhood features for human action recognition (2010, background): “Furthermore, the cumulative nature of the histogram means that closer interest points have more influence on the feature, and we can mitigate sensitivity to exact relative ordering of the neighborhood, as has been shown useful for related features in static images [17, 16].”
Multi-View K-Means Clustering on Big Data (2013, unknown): “Following [Lee and Grauman, 2009], we select 7 classes composed of tree, building, airplane, cow, face, car, bicycle and each class has 30 images.”
Shape discovery from unlabeled image collections (2009, result): “Our results on benchmark datasets demonstrate the approach can successfully discover shapes from unlabeled images.”
Unsupervised object class discovery via saliency-guided multiple class learning (2012, methodology): “They are either restricted to only categorization (no object localization) [22, 18, 38], or have to resort to a separate detector training process using their localization results [15, 19, 21], or only obtain specialized detectors such as chamfer distance based shape templates [20].”
Discovering localized attributes for fine-grained recognition (2012, unknown): “However, these methods aim to ﬁnd objects that are common across images, while we are interested in ﬁnding discriminative local regions that will maximize classiﬁcation performance.”
Learning the easy things first: Self-paced visual category discovery (2011, methodology): “Unsupervised visual discovery methods detect appearance patterns using either clustering algorithms [10, 6, 11] or latent topic models [19, 13].”
From a Set of Shapes to Object Discovery (2010, methodology): “3. For all three graphs, we apply the standard PageRank algorithm, also used in [9, 17 , 10], to identify the most relevant contours, which are then interpreted as object contours.”
Avoiding Confusing Features in Place Recognition (2010, methodology): “Finally, the task of confuser detection has some similarities with the task of feature selection in category-level recognition [25,26,27] and retrieval [28,29,30].”
Weakly-supervised Discovery of Visual Pattern Configurations (2014, background): “Modeling the spatial and geometric relationship between co-occurring visual patterns (objects or object-parts) often improves visual recognition performance [30, 9, 25, 21, 31, 16, 8, 6, 22, 17].”