The SUN Attribute Database: Beyond Categories for Deeper Scene Understanding

By Genevieve Patterson, Chen Xu, Hang Su, James Hays. 2014

424 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

How researchers see this paper

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

In Artificial Intelligence

In Computer Vision and Pattern Recognition

Citations in the corpus

Joint Discovery of Object States and Manipulation Actions (2017, background): “Prior work has addressed recognition of object attributes [15, 33, 34], which can be seen as different object states in some cases.”
Learning Representations for Automatic Colorization (2016, methodology): “We do not use any SUN-A images for training.”
Examining CNN representations with respect to Dataset Bias (2017, methodology): “Dataset: We tested the proposed method on the Large-scale CelebFaces Attributes (CelebA) dataset (Liu et al. 2015) and the SUN Attribute database (Patterson et al. 2014).”
Automatic Description Generation from Images: A Survey of Models, Datasets, and Evaluation Measures (2016, methodology): “Patterson et al. (2014) were the first to present a large-scale scene attribute dataset in the computer vision community.”
On the relationship between visual attributes and convolutional networks (2015, neutral): “Using large datasets of attributes and scenes [14, 24], we can extend the applicability of our work, as well as, incorporate other forms of regularization to encourage other priors such as shared activations or correlations between attributes.”
A Review of Generalized Zero-Shot Learning Methods (2020, background): “n of AWA1 that includes more images. UNderstanding (SUN) [184] data set, which is a well-known scene-based data set, contains 130,519 images in 899 classes (categories). SUN attribute data set [185], [186] is a subset of the original SUN data set for ﬁne-grained scene classiﬁcation (categorization).”
Normalization Matters in Zero-Shot Learning (2020, methodology): “We test our approach in CZSL scenario on two datasets: CUB [46] and SUN [42].”
Improving Semantic Embedding Consistency by Metric Learning for Zero-Shot Classiffication (2016, result): “…datasets widely used in the community, allowing to compare our results with those recently proposed in the literature: the aPascal&aYahoo (aP&Y) [23], Animals with Attributes (AwA) [3], CUB-200-2011 (CUB) [44] and SUN attribute (SUN) [45] datasets (see Table 1 for few statistics on their content).”
Building a Large-scale Multimodal Knowledge Base System for Answering Visual Queries (2015, methodology): “We use the SUN Attribute Dataset [37], which provides 102 attribute labels (e.g., glossy and warm).”
From BoW to CNN: Two Decades of Texture Representation for Texture Classification (2018, background): “The study of visual texture attributes [17, 34, 151] was motivated by the signiﬁcant interest raised by visual attributes [58, 176, 175, 105].”
DeepFashion: Powering Robust Clothes Recognition and Retrieval with Rich Annotations (2016, methodology): “Extensive experiments demonstrate the effectiveness of FashionNet and the usefulness of DeepFashion.”
Nonparametric Method for Data-driven Image Captioning (2014, methodology): “Follow-up work (Patterson et al., 2014) shows that their attributes provide improved matching for image captioning over I M 2T EXT baseline.”
Attribute Prototype Network for Zero-Shot Learning (2020, methodology): “We directly adopt the disjoint attribute groups defined by the datasets [39, 20, 31].”
Zero-Shot Learning via Joint Latent Similarity Embedding (2015, background): “We test our method on four benchmark image datasets for zero-shot recognition and retrieval, i.e. aPascal & aYahoo (aP&Y) [10], Animals with Attributes (AwA) [17], Caltech-UCSD Birds-200-2011 (CUB-200-2011) [36], and SUN Attribute [29].”
Zero-Shot Recognition via Structured Prediction (2016, methodology): “Empirically we test our method on four popular benchmark image datasets for ZSL, namely, aPascal & aYahoo (aP&Y) [3], Animals with Attributes (AwA) [18], Caltech-UCSD Birds-200-2011 (CUB) [19], and SUN Attribute (SUN) [20], and achieve the state-of-the-art.”
Domain-Invariant Projection Learning for Zero-Shot Recognition (2018, background): “Four of them are of medium-size: Animals with Attributes (AwA) [25], CUB-200-2011 Birds (CUB) [56], aPas-cal&Yahoo (aPY) [13], and SUN Attribute (SUN) [41].”