Embodied Visual Object Recognition
Marcus Wallenberg · Linköping studies in science and technology. Dissertations · 2016
Object recognition is a skill we as humans often take for granted.Due to our formidable object learning, recognition and generalisation skills, it is sometimes hard to see the multitude of obstacles that need to be overcome in order to replicate this skill in an artificial system.Object recognition is also one of the classical areas of computer vision, and many ways of approaching the problem have been proposed.Recently, visually capable robots and autonomous vehicles have increased the focus on embodied recognition systems and active visual search.These applications demand that systems can learn and adapt to their surroundings and arrive at decisions in a reasonable amount of time.Ideally, this should be done while maintaining high object recognition performance.This is especially challenging due to the high dimensionality of image data and in cases where end-to-end learning from pixels to output is needed.Therefore, mechanisms designed to make inputs tractable are often necessary for less computationally capable embodied systems.Active visual search also means that mechanisms for attention and gaze control are integral to the object recognition procedure.Therefore, the way in which attention mechanisms should be introduced into feature extraction and estimation algorithms must be carefully considered when constructing a recognition system.This thesis describes work done on the components necessary for creating an embodied recognition system, specifically in the areas of decision uncertainty estimation, object segmentation from multiple cues, adaptation of stereo vision to a specific platform and setting, problem-specific feature selection, efficient estimator training and attentional modulation in convolutional neural networks.Contributions include the evaluation of methods and measures for predicting the potential uncertainty reduction that can be obtained from additional views of an object, allowing for adaptive target observations.Also, in order to separate a specific object from other parts of a scene, it is often necessary to combine multiple cues such as colour and depth in order to obtain satisfactory results.Therefore, a method for combining these using channel coding has been evaluated.In order to make use of three-dimensional spatial structure in recognition, a novel stereo vision algorithm extension along with a framework for automatic stereo tuning have also been investigated.Feature selection and efficient discriminant sampling for decision tree-based estimators have also been implemented.Finally, attentional multi-layer modulation of convolutional neural networks for recognition in cluttered scenes has been investigated.Several of these components have been tested and evaluated on a purpose-built embodied recognition platform known as Eddie the Embodied.v At a recent PopSci event, I was approached by one of the other exhibitors and told that what had tipped the scales and made her go into engineering was a demonstration of Eddie at a similar event several years prior.Likewise, my own endeavours into computer vision research have been inspired and influenced by a great many people in a great many ways.From professional and scientific advice to inspiration for guitar riffs and, at one time, the donation of an organ 1 , these have all played their parts in bringing this work and its author to their current state.Although too numerous to mention individually, I would like to extend my special thanks to:• My supervisors Per-Erik Forssén and Michael Felsberg, for giving me the trust and opportunity to carry out this work, and for their continued support throughout.• Everyone at CVL (past and present) whose knowledge and ideas were vital in providing insight, support and inspiration.• My wife Lisa, for not letting me give up despite professional and personal setbacks, and my own serious doubts about the outcome of this work.• My parents