In many real-world tasks of image classification, limited amounts of labeled data are available to train automatic classifiers. Consequently, extensive human expert involvement is required for verification. A novel solution is presented that makes use of active learning combined with an ensemble of classifiers for each class. The result is a significant reduction in required expert involvement ...

Smooth control using an active vision head’s verge-axis joint is performed through continuous state and action reinforcement learning. The system learns to perform visual servoing based on rewards given relative to tracking performance. The learned controller compensates for the velocity of the target and performs lag-free pursuit of a swinging target. By comparing controllers exposed to differ...

Log-polar mapping plays a fundamental role in active vision by allowing both a wide eld-of-view and a high-accuity area. While log-polar mapping simpli es certain visual tasks, translation estimation becomes more diÆcult in this domain. However, given the importance of translational motion for mobile target tracking, solving this problem in log-polar space is of great signi cance. In this paper...

In many real-world tasks of image classification, limited amounts of labeled data are available to train automatic classifiers. Consequently, extensive human expert involvement is required for verification. A novel solution is presented that makes use of active learning combined with an ensemble of classifiers for each class. The result is a significant reduction in required expert involvement ...

In this paper, we describe methods to localize a mobile robot in an indoor environment from visual information. An appearance-based approach is adopted in which the environment is represented by a large set of images from which features are extracted. We extended the appearance based approach with an active vision component, which fits well in our probabilistic framework. We also describe anoth...

Guangcong Zhang School of Electrical & Computer Engineering, Institute for Robotics & Intelligent Machines, Georgia Institute of Technology, Atlanta, Georgia 30332 e-mail: [email protected] Michail Kontitsis School of Aerospace Engineering, Institute for Robotics & Intelligent Machines, Georgia Institute of Technology, Atlanta, Georgia 30332 e-mail: [email protected] Nuno Filipe School of A...

In recent decades, active vision systems have been studied widely because they can control the gaze direction of their mounted vision sensors and overcome many shortcomings of passive vision systems which have fixed vision sensors. Due to an increasing interest in humanoid robots and social robots, many researchers in the active vision area are focusing their research on developing active visio...

This paper presents recent results from the Robotics Systems Laboratory at the Australian National University in the area of Robot Vision. Progress has been made in the design and construction of novel mechanisms for both active and panoramic vision, together with computer systems that facilitate real-time vision processing. The robot vision systems are being used to develop human-friendly robo...

As an important data selection schema, active learning emerges as the essential component when iterating Artificial Intelligence (AI) model. It becomes even more critical given dominance of deep neural network based models, which are composed a large number parameters and hungry, in application. Despite its indispensable role for developing AI research on is not intensive other directions. In t...