Capturing Long-Tail Distributions of Object Subcategories

By Xiangxin Zhu, Dragomir Anguelov, Deva Ramanan. 2014

179 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

Conceptual 12M: Pushing Web-Scale Image-Text Pre-Training To Recognize Long-Tail Visual Concepts (2021, background): “Long-tail Visual Recognition in V+L Addressing long-tail distributions of visual concepts is an important component of V+L systems that generalize, as long and free-form texts exhibit a large number of compositional, ﬁne-grained categories [84, 51, 19].”
Greedy AutoAugment (2019, methodology): “This is a classic type of long-tail distribution data, which is also prevalent in natural images (Zhu et al., 2014; Tang et al., 2015) and can be addressed by data augmentation methods (Wang et al.”
The More You Know: Using Knowledge Graphs for Image Classification (2016, methodology): “While some visual concepts are very common such as person or car, most categories have many fewer examples, forming a long-tail distribution [37].”
Imbalanced Continual Learning with Partitioning Reservoir Sampling (2020, background): “Imbalanced data induce severe issues in learning that are primarily attributed to gradient dominance and under-representation of the minority [44,21,98,80].”
Incremental Meta-Learning via Indirect Discriminant Alignment (2020, background): “The natural occurrence of classes in the world is a long tailed distribution [42, 37, 39], whereby instances for most classes are rare and instances for few classes are abundant.”
Large-Scale Long-Tailed Recognition in an Open World (2019, background): “Arising from long-tail distributions of natural data, it has been extensively studied [41, 61, 3, 30, 62, 34, 29, 49, 6].”
What Neural Networks Memorize and Why: Discovering the Long Tail via Influence Estimation (2020, background): “While it has been recognized in many contexts that modern datasets are long-tailed [32, 28, 3], it is unclear whether this has any relationship to memorization by modern deep learning algorithms and (if so) how significant is the effect.”
The Devil is in the Tails: Fine-grained Classification in the Wild (2017, background): “In a nutshell: the world is long-tailed, as previously noted in the context of subcategories and object views [43, 64].”
An Empirical Study and Analysis of Generalized Zero-Shot Learning for Object Recognition in the Wild (2016, background): “It is well-known, however, that object frequencies in natural images follow longtailed distributions [1,2,3].”
Large Scale Fine-Grained Categorization and Domain-Specific Transfer Learning (2018, background): “Another issue we address in this paper is the long-tailed distribution, where a few categories have most of the images [71, 56].”
Characterising Bias in Compressed Models (2020, background): “As seen in Figure 1, CelebA is representative of many natural image datasets where attributes follow a long-tail distribution (Zhu et al., 2014; Feldman, 2019).”
Recent Advances in Open Set Recognition: A Survey (2018, background): “Further, as the object frequencies in visual categories ordinarily follow long-tailed distribution [29], [116], such a distribution fitting will face challenges once the rare classes in KKCs and UUCs appear together in testing [117].”
Meta-Learning to Detect Rare Objects (2019, background): “long-tailed, imbalanced datasets, in which a few dominant (‘head’) classes claim most of the examples, while most of the other (‘tail’) classes are represented by relatively few examples [74, 71, 9].”
Predicting Visual Exemplars of Unseen Classes for Zero-Shot Learning (2016, background): “In other words, the frequencies of observing objects follow a long-tailed distribution [37, 51].”
When is memorization of irrelevant training data necessary for high-accuracy learning? (2020, background): “While the problems we describe are intentionally simplified to allow for clean theoretical analysis, they rely on properties found in natural learning problems such as clustering of examples, noise, and a fine-grained subpopulation structure [38].”
Jointly Optimize Data Augmentation and Network Training: Adversarial Data Augmentation in Human Pose Estimation (2018, background): “Second, natural images usually follow long-tail distributions [46, 33].”
REVISE: A Tool for Measuring and Mitigating Bias in Visual Datasets (2020, background): “…et al., 2016; Salakhutdinov et al., 2011; J. Yang et al., 2014), object contexts (Choi et al., 2012; Rosenfeld et al., 2018), object sub-categories (Zhu et al., 2014), scenes (Zhou et al., 2017), gender (Kay et al., 2015), gender contexts (Burns et al., 2018; J. Zhao et al., 2017), skin tones…”
A-OKVQA: A Benchmark for Visual Question Answering using World Knowledge (2022, background): “What we see is a fairly typical long-tail distribution of labels, as is seen in many openlabeled image tasks [55].”
A Two-Stage Approach to Few-Shot Learning for Image Recognition (2019, unknown): “object categories [5] as shown in Fig.”