Object class recognition by unsupervised scale-invariant learning

By Rob Fergus, Pietro Perona, Andrew Zisserman. 2003

2,542 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Media Technology

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

In Media Technology

Citations in the corpus

Unsupervised Learning of Visual Representations by Solving Jigsaw Puzzles (2016, background): “th high-dimensional manifolds). 4 M. Noroozi and P. Favaro In the context of Computer Vision, it is worth mentioning some early work on unsupervised learning of models for classication. For instance [13,40] introduced methods to build a probabilistic representation of objects as constellations of pa”
Learning Generative Visual Models from Few Training Examples: An Incremental Bayesian Approach Tested on 101 Object Categories (2004, methodology): “Progress on this topic has been recently reported by Weber et al. [5] who proposed and demonstrated a method for training in clutter without supervision (see also the follow-up paper by Fergus et al. [8] proposing an improved scale-invariant algorithm).”
LabelMe: A Database and Web-Based Tool for Image Annotation (2008, background): “Research in object detection and recognition would benefit from large image and videocollections with ground truth labels spanning many different object categories in cluttered scenes.”
Unsupervised Discovery of Mid-Level Discriminative Patches (2012, neutral): “[23] had “unsupervised” in its title, it would now be considered a weakly supervised method.”
Using Multiple Segmentations to Discover Objects and their Extent in Image Collections (2006, methodology): “Caltech [9] 4,090 4 + background MSRC [27] 4,325 23 object and scene categories LabelMe [18] 1,554 cars, buildings, trees”
Neural Activation Constellations: Unsupervised Part Model Discovery with Convolutional Networks (2015, methodology): “Although creating a model for the spatial relationship of parts has already been introduced a decade ago [16, 14], these approaches face major difficulties due to the fact that part proposals are based on hand-engineered local descriptors and detectors without correspondence We overcome this problem”
Discovering objects and their location in images (2005, methodology): “This is in contrast to [8], where each image is labelled with an identity of the object it contains, i.”
Context as Supervisory Signal: Discovering Objects with Predictable Context (2014, background): “Proponents of unsupervised representation learning [1,2,3,4] and unsupervised object discovery [5,6,7,8,9,10,11,12] have long argued that these approaches have the potential to solve two fundamental problems with supervised methods.”
Sharing Visual Features for Multiclass and Multiview Object Detection (2007, background): “Alternatively, one can extract local windows at locations and scales returned by an interest point detector and classify these, either as an object or as part of an object (see, e.g., [12]).”
Semi-Supervised Self-Training of Object Detection Models (2005, methodology): “Recent work [8], [7] utilizes an EM based approach.”
CliqueCNN: Deep Unsupervised Exemplar Learning (2016, background): “s the foundation for numerous computer vision subtasks ranging from low-level image processing to high-level object recognition or posture analysis. A common paradigm following the inﬂuential work of [14] has been category-level recognition, where categories and the similarities of all their instanc”
Learning part-based templates from large collections of 3D shapes (2013, methodology): “[2005] and others have developed part-based models to encode distributions of appearances and spatial arrangements of parts and used them for recognition of objects in images [Fergus et al. 2003; Amit and Trouve 2007; Felzenszwalb et al. 2010].”
A codebook-free and annotation-free approach for fine-grained image categorization (2012, background): “[12, 30, 11]) perform well in classifying objects of minimal articulation or localizing objects from the background.”
Search-and-replace editing for personal photo collections (2010, background): “Since the image edits in our application are associated with particular people or objects, we require less flexible matching than for recognition across object class [12].”
Shape Classification Using the Inner-Distance (2007, background): “The first category (e.g., [3], [19], [14], [37], [15], [46], etc.) builds part models from a set of sample images and usually with some prior knowledge such as the number of parts.”
Learning the Compositional Nature of Visual Objects (2007, methodology): “In contrast to modeling an object directly based on a constellation of its parts (e.g. [8]), the compositional approach learns intermediate groupings of parts.”
An Affine Invariant Salient Region Detector (2004, background): “The selection of a set of image regions forms the first step in many computer vision algorithms, for example for computing image correspondences [2,17,19,20, 22], or for learning object categories [1,3,4,23].”