Weak Hypotheses and Boosting for Generic Object Detection and Recognition

By Andreas Opelt, Michael Fussenegger, Axel Pinz, Peter Auer. European Conference on Computer Vision, 2004

367 citations recorded by the source corpus.

Subfields: Aerospace Engineering, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/0c91808994a250d7be332400a534a9291ca3b60e

How researchers see this paper

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

In Aerospace Engineering

In Computer Vision and Pattern Recognition

Citations in the corpus

A Boundary-Fragment-Model for Object Detection (2006, methodology): “[22] simply use a “bag of visual words” model (with no geometrical relations between the parts at all), Agarwal & Roth [1], Amores et al.”
A sparse object category model for efficient learning and exhaustive recognition (2005, background): “These models range from bag-of-word models (where the words are vector quantized invariant descriptors) with no spatial organization [4] [16], through to fragment based models [2] [10] [13] with particular spatial configurations .”
Spatial pyramid co-occurrence for image classification (2011, methodology): “While our primary objective is analyzing overhead imagery, we demonstrate that our approach achieves stateof-the-art performance on the Graz-01 object class dataset and performs competitively on the 15 Scene dataset.”
Spatial pyramid matching (2009, neutral): “To test how well spatial pyramid matching performs under highly variable conditions, we have performed another set of experiments on the Graz dataset (Opelt et al., 2004), which features people and bikes in varying positions and at varying scales, against heavily cluttered backgrounds.”
In defense of Nearest-Neighbor based image classification (2008, methodology): “We compare performance of NBNN against [16, 24, 31].”
Keywords to visual categories: Multiple-instance learning forweakly supervised object categorization (2008, background): “By incorporating constraints that reflect the expected sparsity of true positive examples into a large-margin objective function, our approach remains accurate even when the available text annotations are imperfect and ambiguous.”
The 2005 PASCAL Visual Object Classes Challenge (2005, background): “In recent years there has been a rapid growth in research, and quite some success, in visual recognition of object classes; examples include [1, 5, 10, 14, 18, 28, 39, 43].”
Learning Object Categories From Internet Image Searches (2010, background): “The background to this work is the following: a wide range of different approaches has been taken to the challenging problem of object category recognitionV[2], [11], [16], [25], [30], [32], [40].”
Respectful cameras: detecting visual markers in real-time to address privacy concerns (2007, methodology): “Approaches to object detection employ statistical classification methods including AdaBoost [51, 40], Neural Networks [17], and Support Vector Machines (SVMs) [41].”
50 Years of object recognition: Directions forward (2013, neutral): “[336]), the LabelMe dataset (Russell et al.”
Weakly Supervised Learning of Part-Based Spatial Models for Visual Object Recognition (2006, background): “The first of these is the Graz bicycle image set [9], consisting of 150 images with bicycles and 150 negative images.”
A Linear Time Histogram Metric for Improved SIFT Matching (2008, methodology): “The SIFT descriptor has proven to be successful in applications such as object recognition [3, 7–9], object class detection [10–12], image retrieval [13–16], robot localization [17], building panoramas [18] and image classiﬁcation [19].”
MILES: Multiple-Instance Learning via Embedded Instance Selection (2006, methodology): “[42] described a boosting approach to object detection where, at each iteration, the weak hypothesis finder selects one local image patch and one type of local descriptor.”
Robust Object Detection with Interleaved Categorization and Segmentation (2008, neutral): “Opelt [ 54 ] 92.2% – Thureson [68] 93.2% – Fergus [23] 93.3% 90.3% J. Zhang [82] 98.5% 98.3% Deselaers [18] – 98.9% W. Zhang [83] 99.0% –”
A performance evaluation of local descriptors (2005, background): “recognition [21], image retrieval [29, 38], robot localization [40], video data mining [41], building panoramas [4], and recognition of object categories [8, 9, 22, 35].”
Generic object recognition with boosting (2006, methodology): “This database 5 (further on termed GRAZ-01) that was used in [30], contains 450 images of category person (P), 350 of category bike (B), and 250 of category " counter-class " (N, meaning it contains no bikes and no persons).”
Mining Mid-level Features for Image Classification (2014, methodology): “For the GRAZ-01 dataset we report ROC equal error rate.”