Towards Scalable Representations of Object Categories: Learning a Hierarchy of Parts

By Sanja Fidler, Ales Leonardis. 2007

223 citations recorded by the source corpus.

Subfields: 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 Computer Vision and Pattern Recognition

Citations in the corpus

DeCAF: A Deep Convolutional Activation Feature for Generic Visual Recognition (2013, background): “Such models have been able to perform better than traditional hand-engineered representations in many domains, especially those where good features have not already been engineered (Le et al., 2011).”
Joint Video and Text Parsing for Understanding Events and Answering Queries (2013, background): “In particular, spatial parsing identiﬁes the spatial compositional structures of objects and scenes [21], [6], [22], [23], [24], [25], [7], [8]; temporal parsing recognizes the temporal compositional structures of events [26], [27], [28], [29], [30], [31], [9], [32]; and causal parsing detects the…”
Scalable multi-class object detection (2011, methodology): “For example, when sharing features among categories, [28, 13] have shown that the number of codebook entries is growing only sublinearly in |C| and since sharing makes better use of the training data, it increases the accuracy.”
A probabilistic model for component-based shape synthesis (2012, methodology): “Our probabilistic model is related to a number of hierarchical generative models for object recognition in images [Bouchard and Triggs 2005; Tu et al. 2005; Jin and Geman 2006; Fidler and Leonardis 2007; Todorovic and Ahuja 2008; Zhu et al. 2008; Ommer and Buhmann 2010; Roux et al. 2011; Ranzato et ”
Learning Hybrid Image Templates (HIT) by Information Projection (2012, background): “Image templates, especially deformable ones, have been extensively studied for detection, recognition, and tracking, for example, deformable templates [9], active appearance models [10], pictorial structures [11], constellation model [12], part-based latent SVM model [13], recursive compositional mo”
A generative vision model that trains with high data efficiency and breaks text-based CAPTCHAs (2017, unknown): “Repeating this process for all the training images accumulates counts on the usage of different features at level k, and the final features for this level are selected by optimizing an objective function that balances compression and reconstruction error (31).”
Hierarchical Convolutional Deep Learning in Computer Vision (2013, background): “Fidler and Leonardis [36, 37] propose a top-down model for object recognition which has an explicit notion of parts whose correspondence is explicitly reasoned about at each level.”
Augmenting deformable part models with irregular-shaped object patches (2012, background): “Part-based object detectors have shown remarkable performance in recent years [1, 2, 3, 5, 6, 7, 21, 25].”
Teaching Compositionality to CNNs (2017, background): “The first flavor focuses on the creation of hierarchical feature representations by means of statistical modeling [8, 27, 48, 47], reusable deformable parts [49, 29], or compositional graph structures [36, 45].”
Learning AND-OR Templates for Object Recognition and Detection (2013, background): “In the literature, deformable templates [3], [7], [6], [20], [5], [22], [21] and compositional hierarchy [11], [12], [9], [23], [19], [16], [8], [2] are widely used in object modeling with shared parts among categories .”
Factorizable Net: An Efficient Subgraph-based Framework for Scene Graph Generation (2018, background): “…relationships or phrases detection and detection works and these works were mostly supposed to leveraging other tasks, such as object recognition [4,12,13,31,32,44,51,53,55], image classiﬁcation and retrieval [17,39], scene understanding and generation [2,3,18,23,24,61,65], as well as text…”
Deep Neural Networks for Object Detection (2013, methodology): “Further examples of compositional models for detection are based on segments as primitives [1], focus on shape [13], use Gabor filters [10] or larger HOG filters [19].”
Learning Semantic Image Representations at a Large Scale (2014, background): “It has long been argued that deep or layered compositional architectures should be able to capture salient aspects of a given domain through discovery of salient clusters, parts, mid-level features, and/or hidden units [48, 41, 125, 102, 65].”
Learning generative models of mid-level structure in natural images (2012, background): “In most cases it is not possible to generate new images from the learned models, and unsupervised learning can be a problem (but see, for instance, Fidler and Leonardis, 2007; Todorovic and Ahuja, 2008).”
Semantic hierarchies for image annotation: A survey (2012, background): “Fidler and Leonardis [37] also build a hierarchy of parts using unsupervised statistics.”