Adaptive Computer Vision: Online Learning for Object Recognition

Bekel, Holger; Bax, Ingo; Heidemann, Gunther; Ritter, Helge

doi:10.1007/978-3-540-28649-3_55

Cited by 16 publications

(23 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For the application to online learning, using only the STM model achieved good generalization in combination with a large storage capacity of 50 objects, compared to other work on online learning of objects which usually did not consider more than 10-12 objects (Bekel, Bax, Heidemann, & Ritter 2004;Arsenio 2004). This capacity is a direct consequence of the high-dimensional representation space, and is also achieved if only shape representations are used.…”

Section: Discussionmentioning

confidence: 88%

“…In the following we discuss the components of our model with reference to related work. Our feature detection approach is different from most of the related work on online learning for object recognition (Garcia, Oliveira, Grupen, Wheeler, & Fagg 2000;Steels & Kaplan 2001;Roy & Pentland 2002;Arsenio 2004;Bekel, Bax, Heidemann, & Ritter 2004), because the representation is not based on a dimension reduction of the high-dimensional visual input. Due to the receptive-field-based topographical representation, we obtain multiple shape feature-map representations with a resulting dimensionality that is of the same order as the visual input.…”

Section: Discussionmentioning

confidence: 99%

“…Due to the problems of standard architectures like MLPs, which suffer from catastrophic forgetting in such a scenario, most previous work on online object learning does not consider incremental learning, but rather collects the training data and then performs a standard batch learning procedure (Bekel, Bax, Heidemann, & Ritter 2004).…”

Section: Discussionmentioning

confidence: 99%

“…An interesting approach to supervised online learning for object recognition was proposed by Bekel, Bax, Heidemann, & Ritter (2004). Their classification architecture consists of three major stages.…”

Section: Online Learning and Man-machine Interactionmentioning

confidence: 99%

See 3 more Smart Citations

A biologically motivated visual memory architecture for online learning of objects

2008

View full text Add to dashboard Cite

We present a biologically motivated architecture for object recognition that is based on a hierarchical feature detection model in combination with a memory architecture that implements short-term and long-term memory for objects. A particular focus is the functional realization of online and incremental learning for the task of appearance-based object recognition of many complex-shaped objects. We propose some modifications of learning vector quantization algorithms that are especially adapted to the task of incremental learning and capable of dealing with the stabilityplasticity dilemma of such learning algorithms. Our technical implementation of the neural architecture is capable of online learning of 50 objects within less than three hours.

show abstract

Section: Discussionmentioning

confidence: 88%

Section: Discussionmentioning

confidence: 99%

Section: Discussionmentioning

confidence: 99%

Section: Online Learning and Man-machine Interactionmentioning

confidence: 99%

See 2 more Smart Citations

A biologically motivated visual memory architecture for online learning of objects

2008

View full text Add to dashboard Cite

show abstract

“…Due to the lack of rapid learning methods for complex shapes, research in man-machine interaction for robotics dealing with online learning of objects has mainly used histogram-based feature representations that offer fast processing [5,1], but only limited representational and discriminatory capacity. An interesting approach to online learning for object recognition was proposed by Bekel et al [2]. Their VPL classifier consists of feature extraction based on vector quantisation and PCA and supervised classification using a local linear map architecture.…”

Section: Introductionmentioning

confidence: 99%

Online Learning for Object Recognition with a Hierarchical Visual Cortex Model

Kirstein

Wersing

Körner

2005

Artificial Neural Networks: Biological Inspirations – ICANN 2005

View full text Add to dashboard Cite

Abstract. We present an architecture for the online learning of object representations based on a visual cortex hierarchy developed earlier. We use the output of a topographical feature hierarchy to provide a viewbased representation of three-dimensional objects as a form of visual short term memory. Objects are represented in an incremental vector quantization model, that selects and stores representative feature maps of object views together with the object label. New views are added to the representation based on their similarity to already stored views. The realized recognition system is a major step towards shape-based immediate high-performance online recognition capability for arbitrary complex-shaped objects.

show abstract

Rapid Online Learning of Objects in a Biologically Motivated Recognition Architecture

Kirstein

Wersing

Körner

2005

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Abstract. We present an approach for the supervised online learning of object representations based on a biologically motivated architecture of visual processing. We use the output of a recently developed topographical feature hierarchy to provide a view-based representation of threedimensional objects using a dynamical vector quantization approach. For a simple short-term object memory model we demonstrate real-time online learning of 50 complex-shaped objects within three hours. Additionally we propose some modifications of learning vector quantization algorithms that are especially adapted to the task of online learning and capable of effectively reducing the representational effort in a transfer from short-term to long-term memory.

show abstract

Adaptive Computer Vision: Online Learning for Object Recognition

Cited by 16 publications

References 13 publications

A biologically motivated visual memory architecture for online learning of objects

A biologically motivated visual memory architecture for online learning of objects

Online Learning for Object Recognition with a Hierarchical Visual Cortex Model

Rapid Online Learning of Objects in a Biologically Motivated Recognition Architecture

Contact Info

Product

Resources

About