What Does Classifying More Than 10, 000 Image Categories Tell Us?

By Jia Deng, Alexander C. Berg, K. Li, Li Fei-Fei. European Conference on Computer Vision, 2010

636 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/9800e3c3394c569be83379ee2ebe3424e09c2919

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

Citations in the corpus

Semantic Kernel Forests from Multiple Taxonomies (2012, methodology): “We demonstrate our method on challenging object recognition datasets, and show that interleaving multiple taxonomic views yields significant accuracy improvements.”
Hedging your bets: Optimizing accuracy-specificity trade-offs in large scale visual recognition (2012, methodology): “We use three datasets ranging from 65 to over 10,000 classes: ILSVRC65, ILSVRC1K [28], and ImageNet10K [7].”
Fine-Grained Image Classification by Exploring Bipartite-Graph Labels (2015, background): “Previous works have shown the benefits of leveraging the semantic class hierarchy using either unsupervised [2, 50] or supervised [11, 18, 39] methods.”
From Large Scale Image Categorization to Entry-Level Categories (2013, background): “Although far from solved, algorithms have now advanced to the point where they can recognize or localize thousands of object categories with reasonable accuracy [17, 4, 3, 12].”
Write a Classifier: Predicting Visual Classifiers from Unstructured Text (2015, background): “For example in [5], it was shown that there is a strong correlation between semantic similarity between classes, based on WordNet, and confusion between classes.”
A codebook-free and annotation-free approach for fine-grained image categorization (2012, background): “There is a number of recent work in fine-grained image classification using various datasets such as flowers [22], stonefly larvae [20], bird species [29], and related leaf nodes of ImageNet [9] such as all fungus [8].”
A hierarchical loss and its problems when classifying non-hierarchically (2017, background): “The particular special cases discussed by them and by [4–9] allocate different weights to different leaves, not leveraging “ultrametric trees” as in the present paper (an ultrametric tree is a tree with a so-called ultrametric distance metric such that all leaves are at the same distance from the ro”
Metric Learning for Large Scale Image Classification: Generalizing to New Classes at Near-Zero Cost (2012, neutral): “The current state-of-the-art [8,10] uses efﬁ-cient linear SVM classiﬁers trained in a one-vs-rest manner in combination with high-dimensional bag-of-word [20] or Fisher vector representations [19].”
The Inverted Multi-Index (2012, background): “image classification [3, 5], understanding [14], image editing [6], etc.”
A Robust 3D-2D Interactive Tool for Scene Segmentation and Annotation (2016, background): “, [12], [13], [14]) various tasks like semantic segmentation, object detection and recognition [15], [16].”
Bayesian Joint Modelling for Object Localisation in Weakly Labelled Images (2015, background): “OBJECT recognition is a challenging problem especially at a large scale because of variabilities in object appearance, viewpoint, illumination and pose [1], [2], [3].”
ImageNet Auto-Annotation with Segmentation Propagation (2014, background): “Several works tackle image classiﬁcation [18,42,21,20,36, 17] or object detection in the fully supervised setting [17].”
Towards a Visual Turing Challenge (2014, background): “Architectures that aim at mimicking or reproducing this space of human conceptsneed to capture the same diversity and therefore scale up to thousands of concepts [38, 39, 40].”
Universal statistics of Fisher information in deep neural networks: mean field approach (2018, background): “However, since classification problems with far more labels are sometimes examined in the context of machine learning [45], it would be helpful to remark on the case of C = O(M) here.”
What do 15,000 object categories tell us about classifying and localizing actions? (2015, neutral): “Instead of localizing cuboids [2], recently action location is more precisely deﬁned as a sequence of bounding boxes [15, 22].”
The Language Demographics of Amazon Mechanical Turk (2014, background): “…to research areas like human-computer interaction (Bigham et al., 2010; Bernstein et al., 2010), computer vision (Sorokin and Forsyth, 2008; Deng et al., 2010; Rashtchian et al., 2010), speech processing (Marge et al., 2010; Lane et al., 2010; Parent and Eskenazi, 2011; Eskenazi et al.,…”