Online Learning for Bootstrapping of Object Recognition and Localization in a Biologically Motivated Architecture

Wersing, Heiko; Kirstein, Stephan; Schneiders, Bernd; Bauer-Wersing, Ute; Körner, Edgar

doi:10.1007/978-3-540-79547-6_37

Cited by 5 publications

(5 citation statements)

References 15 publications

(27 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For this reason, depth information has been exploited in a variety of robotics applications in the past [5], [6], [7], [8], [9], [10]. However, it is not easy to find methods for depth estimation from a stereo pair which are a good trade-off between robustness (e.g.…”

Section: Related Workmentioning

confidence: 99%

Enabling Depth-Driven Visual Attention on the iCub Humanoid Robot: Instructions for Use and New Perspectives

et al. 2016

View full text Add to dashboard Cite

Reliable depth perception eases and enables a large variety of attentional and interactive behaviors on humanoid robots. However, the use of depth in real-world scenarios is hindered by the difficulty of computing real-time and robust binocular disparity maps from moving stereo cameras. On the iCub humanoid robot, we recently adopted the Efficient Large-scale Stereo (ELAS) Matching algorithm (Geiger et al., 2010) for computation of the disparity map. In this technical report, we show that this algorithm allows reliable depth perception and experimental evidence that demonstrates that it can be used to solve challenging visual tasks in real-world indoor settings. As a case study, we consider the common situation where the robot is asked to focus the attention on one object close in the scene, showing how a simple but effective disparity-based segmentation solves the problem in this case. This example paves the way to a variety of other similar applications.

show abstract

Section: Related Workmentioning

confidence: 99%

Enabling Depth-Driven Visual Attention on the iCub Humanoid Robot: Instructions for Use and New Perspectives

et al. 2016

View full text Add to dashboard Cite

show abstract

“…As a first step, a hierarchical feed-forward network is applied to the camera image in the manner of a convolutional network [22]. This produces a pyramid of K retinotopic confidence, or, if we wish to stay in the language of Bayesian inference, object likelihood maps.…”

Section: Signal-driven Object Detectionmentioning

confidence: 99%

“…In general, this identity estimate will be error-prone; however, if it is correct "on average", it is intuitive that successful training of context models could be feasible. To assess this, the experiments of this article are conducted in two conditions which differ in the way the signal-driven detection algorithm of [22] is trained. In the default condition, training is done using ground-truth vehicle data taken from the HRI RoadTraffic dataset [7].…”

Section: Introductionmentioning

confidence: 99%

Co-training of context models for real-time vehicle detection

Gepperth

2012

2012 IEEE Intelligent Vehicles Symposium

View full text Add to dashboard Cite

We describe a simple way to reduce the amount of required training data in context-based models of realtime object detection. We demonstrate the feasibility of our approach in a very challenging vehicle detection scenario comprising multiple weather, environment and light conditions such as rain, snow and darkness (night). The investigation is based on a real-time detection system effectively composed of two trainable components: an exhaustive multiscale object detector ("signal-driven detection"), as well as a module for generating object-specific visual attention ("context models") controlling the signal-driven detection process. Both parts of the system require a significant amount of ground-truth data which need to be generated by human annotation in a time-consuming and costly process.Assuming sufficient training examples for signal-based detection, we demonstrate that a co-training step can eliminate the need for separate ground-truth data to train context models. This is achieved by directly training context models with the results of signal-driven detection. We show that this process is feasible for different qualities of signal-driven detection, and maintains the performance gains from context models.As it is by now widely accepted that signal-driven object detection can be significantly improved by context models, our method allows to train strongly improved detection systems without additional labor, and above all, cost.

show abstract

“…The SLP learns to generate high values for positive examples and low values for negative ones. The result is a so called viewtuned-unit [20] which responds robustly to car views of different viewing angle, delivering competitive performance in benchmarks like the UINC car detection [21].…”

Section: B Appearance-based Classifiermentioning

confidence: 99%

System approach for multi-purpose representations of traffic scene elements

Schmuedderich

Einecke

Hasler

et al. 2010

13th International IEEE Conference on Intelligent Transportation Systems

Self Cite

View full text Add to dashboard Cite

Abstract-A major step towards intelligent vehicles lies in the acquisition of an environmental representation of sufficient generality to serve as the basis for a multitude of different assistance-relevant tasks. This acquisition process must reliably cope with the variety of environmental changes inherent to traffic environments. As a step towards this goal, we present our most recent integrated system performing object detection in challenging environments (e.g., inner-city or heavy rain). The system integrates unspecific and vehicle-specific methods for the detection of traffic scene elements, thus creating multiple object hypotheses. Each detection method is modulated by optimized models of typical scene context features which are used to enhance and suppress hypotheses. A multi-object tracking and fusion process is applied to make the produced hypotheses spatially and temporally coherent. In extensive evaluations we show that the presented system successfully analyzes scene elements under diverse conditions, including challenging weather and changing scenarios. We demonstrate that the used generic hypothesis representations allow successful application to a variety of tasks including object detection, movement estimation, and risk assessment by time-to-contact evaluation.

show abstract

Online Learning for Bootstrapping of Object Recognition and Localization in a Biologically Motivated Architecture

Cited by 5 publications

References 15 publications

Enabling Depth-Driven Visual Attention on the iCub Humanoid Robot: Instructions for Use and New Perspectives

Enabling Depth-Driven Visual Attention on the iCub Humanoid Robot: Instructions for Use and New Perspectives

Co-training of context models for real-time vehicle detection

System approach for multi-purpose representations of traffic scene elements

Contact Info

Product

Resources

About