A sparse object category model for efficient learning and exhaustive recognition

By Rob Fergus, Pietro Perona, Andrew Zisserman. Computer Vision and Pattern Recognition, 2005

343 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/162065fb9de1928f7abd593ee9a1b7d41b5a4310

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

50 Years of object recognition: Directions forward (2013, background): “Fergus, Perona and Zisserman are arguably some of the strongest advocates of the approach and have published a series of papers on constellation methods, some of which we overview here [100], [195], [174], [200], [203].”
An Integrated Model of Top-Down and Bottom-Up Attention for Optimizing Detection Speed (2006, background): “Role of attention in accelerating detection speed: Recent models of object detection overcome the speed bottleneck of the sliding window approach by using a generic attention operator to quickly select a few interest points in the image [7, 2].”
Weakly Supervised Learning of Part-Based Spatial Models for Visual Object Recognition (2006, methodology): “method produces better results than previous techniques that use a fixed set of feature detectors rather than simultaneously learning part appearance and spatial models [6, 7].”
Object Detection with Discriminatively Trained Part Based Models (2010, background): “Tree (and star) structured pictorial structure models [9], [15], [19] allow for the use of dynamic programming and generalized distance transforms to efficiently search over all possible object configurations in an image, without restricting the possible locations for each part.”
A Boundary-Fragment-Model for Object Detection (2006, neutral): “BFM-D [18] BFM-C [12] [22] [25] [2] [3] [14] [26] [28]”
Features for image retrieval: an experimental comparison (2008, neutral): “[40, 61, 62].”
Large scale mining and retrieval of visual data in a multimodal context (2008, background): “thus suggest a simplified and more efficient star topology in [Fergus et al., 2005].”
Birdlets: Subordinate categorization using volumetric primitives and pose-normalized appearance (2011, background): “Constellation models [11] [48] and their computationally attractive variants [14] [27] are composed of a set of local part detectors together with one or more probability density functions describing the parts' relative locations.”
Example Based 3D Reconstruction from Single 2D Images (2006, methodology): “5.1). To make this norm robust to illumination changes we normalize the intensities in each window to have zero mean and unit variance, similarly to the normalization often applied to patches in detection and recognition methods (e.g., [ 11 ]).”
Feature representation for statistical-learning-based object detection: A review (2015, methodology): “The normalized gradients with PCA projection have been added for object region description in [99].”
Adaptive pose priors for pictorial structures (2010, background): “However, a common problem in such models is poor localization of parts that have weak appearance cues or are easily confused with background clutter (for example, state-of-the-art accuracy for lower arms in human ﬁgures is almost half of that for torso or head [1]).”
Robust Object Detection with Interleaved Categorization and Segmentation (2008, background): “In later versions, Fergus et al. (2005) therefore replaced the fully-connected graph by a simpler star topology, which can handle a far larger number of parts using efﬁcient inference algorithms (Felzenszwalb and Huttenlocher 2005).”
OPTIMOL: Automatic Online Picture Collection via Incremental Model Learning (2007, background): “Due to the vast number of object categorization approaches (Fei-Fei et al. 2007; Felzenszwalb and Huttenlocher 2005; Fergus et al. 2005a, 2005b ; K rempp et al.2002; LeCun et al. 2004; Leibe and Schiele 2004 ;L owe1999), it is out of the scope of our paper to discuss all of them.”
A Comparison between Recursive Neural Networks and Graph Neural Networks (2006, methodology): “EXPERIMENTAL RESULTS RNNs and GNNs have been experimentally compared on an image classification problem which was based on images extracted from the Caltech benchmark database [35].”
Fast Keypoint Recognition in Ten Lines of Code (2007, background): “They range from image registration to object detection [7, 6] and often require real-time performance.”
Image retrieval: Ideas, influences, and trends of the new age (2008, background): “As a compromise, sparser topologies have been proposed, such as the star topology [Fergus et al. 2005], a hierarchy with the lowest levels corresponding to local features [Bouchard and Triggs 2005], and a geometry where local features are spatially dependent on their nearest neighbors [Carneiro…”
One-shot learning of object categories (2006, methodology): “Imposing conditional independence by the use of a tree-structured model would reduce the complexity to OðN(2)P Þ in learning and OðNP Þ in recognition [11], [15].”
Local features are not lonely – Laplacian sparse coding for image classification (2010, methodology): “Many image representation models have been proposed for this problem, such as Part-based model [3], Bag of Words(BoW) model [18], etc.”