3 views
Seen by:Towards recognizing “cool”: Can end users help computer vision recognize subjective attributes of objects in images?
by Todd Kulesza
Published in the proceedings of IUI 2012.
Recent computer vision approaches are aimed at richer image interpretations that extend the standard recognition of... more Recent computer vision approaches are aimed at richer image interpretations that extend the standard recognition of objects in images (e.g., cars) to also recognize object attributes (e.g., cylindrical, has-stripes, wet). However, the more idiosyncratic and abstract the notion of an object attribute (e.g., "cool" car), the more challenging the task of attribute recognition. This paper considers whether end users can help vision algorithms recognize highly idiosyncratic attributes, referred to here as subjective attributes. We empirically investigated how end users recognized three subjective attributes of cars—"cool", "cute", and "classic". Our results suggest the feasibility of vision algorithms recognizing subjective attributes of objects, but an interactive approach beyond standard supervised learning from labeled training examples is needed.
3 views
Seen by:La aplicación de las nuevas tecnologías de visión cimputacional en el registro y la modelización de yacimientos arqueológicos
Trabajo de investigación de tercer ciclo del Doctorado en Arqueología prehistórica. Director: Joan A. Barceló
Polarised light stress analysis and laser scatter imaging for non-contact inspection of heat seals in food trays
Pre-publication paper co-authored with Mike Dudbridge and Tom Duckett
This paper introduces novel non-contact methods for detecting faults in heat seals of food packages. Two alternative... more This paper introduces novel non-contact methods for detecting faults in heat seals of food packages. Two alternative imaging technologies are investigated; laser scatter imaging and polarised light stress images. After segmenting the seal area from the rest of the respective image, a classifier is trained to detect faults in different regions of the seal area using features extracted from the pixels in the respective region. A very large set of candidate features, based on statistical information relating to the colour and texture of each region, is first extracted. Then an adaptive boosting algorithm (AdaBoost) is used to automatically select the best features for discriminating faults from non-faults. With this approach, different features can be selected and optimised for the different imaging methods. In experiments we compare the performance of classifiers trained using features extracted from laser scatter images only, polarised light stress images only, and a combination of both image types. The results show that the polarised light and laser scatter classifiers achieved accuracies of 96% and 90%, respectively, while the combination of both sensors achieved an accuracy of 95%. These figures suggest that both systems have potential for commercial development.
Dragonfly: An Ecological Approach to Digital Architectural Design
Published in ACADIA 2011: Integration Through Computation, ed. by J.M. Taron, V. Parlac, B. Kolarevic and J.S. Johnson, pp.178-186. Stroughton, WI: The Printing House, 2011.
(Co-authored with Daniel Hambleton)
In his keynote address delivered to The American Society for Esthetics in 1976, James J. Gibson wrote, “Architecture... more
In his keynote address delivered to The American Society for Esthetics in 1976, James J. Gibson wrote, “Architecture and design do not have a satisfactory theoretical basis.” He then asked, “Can an ecological approach to the psychology of perception and behavior provide it?” (1976, p. 413) We believe that it can, at least in part. In this paper, we expand upon Gibson’s insights into the nature of perceptual experience by applying the concept of “affordances” to the design of architectural objects in general, and to the domain of digital architectural design in particular. On our account, the affordance-concept supplies a useful theoretical basis for conceptualizing the relationship between environments and occupants with respect to the form and behavioral meaning of geometrically constructed layouts.
Donald Norman (1988) first introduced affordances to interaction design theorists, as a conceptual tool for predicting how agents will interact with a given product. The extensive body of literature that has since emerged, from human-computer-interaction studies (Ackerman, 1996; Conn, 1995; Moran, 1997; Norman, 1999) to architectural theory and practice (Koutamanis, 2006; Maier and Fadel, 2009), has followed Norman’s lead in defining affordances, somewhat amorphously, as whichever action-related properties of objects are sufficient to elicit the intended forms of behavioral interaction between the agent and object. However, while this is correct, it is only half the story. It leaves unexplained how human perceivers detect and “pair down” on the potentially vast range of possible affordances (at a given time), to select the ones that will be relevant to the coordination and guidance of the targeted actions. Call this the “selectivity problem,” a proper treatment of which is missing from the literature. This is no small matter. If the theory of affordances is to be useful to architects and designers, if it is to have explanatory and predictive power over how perceivers will interact with their surroundings, then some account of the cognitive procedure by which affordances are selected for the deployment of specific behaviors is necessary. Otherwise, it is unclear what the theory hopes to predict or explain.
To this end, we maintain that the couching of affordances in a framework of human intentionality is not only consistent with Gibson’s theoretical views (i.e., the action-oriented definition of the concept of affordances not only suggests an intentional perspective), indeed, such a perspective is necessary if we are to succeed in implementing the affordance-concept into an architectural design context in a way that addresses the selectivity problem. This is one of the goals of “Dragonfly,” a first attempt at implementing the affordance-based control of perceptually guided-action into a digital design simulation. Dragonfly enables human interaction with geometry by encoding the basic principles of ecological psychology (including a rudimentary form of intentionality) into an interactive CAD environment. New vistas for future research and interdisciplinary approaches to design are then discussed, with a special emphasis on their applicability to architecture.
Prediction as a candidate for learning deep hierarchical models of data
Recent findings [HOT06] have made possible the learning of deep layered hier- archical representations of data... more
Recent findings [HOT06] have made possible the learning of deep layered hier- archical representations of data mimicking the brains working. It is hoped that this paradigm will unlock some of the power of the brain and lead to advances towards true AI.
In this thesis I implement and evaluate state-of-the-art deep learning models and using these as building blocks I investigate the hypothesis that predicting the time-to-time sensory input is a good learning objective. I introduce the Predictive Encoder (PE) and show that a simple non-regularized learning rule, minimizing prediction error on natural video patches leads to receptive fields similar to those found in Macaque monkey visual area V1. I scale this model to video of natural scenes by introducing the Convolutional Predictive Encoder (CPE) and show similar results. Both models can be used in deep architectures as a deep learning module.
3 views
Seen by:Jensen-Bregman LogDet Divergence for Efficient Similarity Computations on Positive Defnite Tensors
Covariance matrices provide an easy platform for fusing multiple features compactly and as a result have found immense... more
Covariance matrices provide an easy platform for fusing multiple features compactly and as a result have found immense success in several computer vision applications, including activity recognition, visual surveillance, and diffusion tensor
imaging. An important task in all of these applications is to compute the distance
between covariance matrices using a (dis)similarity function, for which the natural
choice is the Riemannian metric corresponding to the manifold inhabited by these
matrices. As this Riemannian manifold is not flat, the dissimilarities should take
into account the curvature of the manifold. As a result such distance computations
tend to slow down, especially when the matrix dimensions are large or gradients
are required. Further, suitability of the metric to enable efficient nearest neighbor
retrieval is an important requirement in the contemporary times of big data analytics. To alleviate these difficulties, this paper proposes a novel dissimilarity measure
for covariances, the Jensen-Bregman LogDet Divergence (JBLD). This divergence
enjoys several desirable theoretical properties, at the same time is computationally less demanding (compared to standard measures). To address the problem
of efficient nearest neighbor retrieval on large covariance datasets, we propose a
metric tree framework using kmeans clustering on JBLD. We demonstrate the superior performance of JBLD on covariance datasets from several computer vision
applications.
'Structure-from-Motion' photogrammetry: a novel, low-cost tool for geomorphological applications
by Matt Westoby
Poster presented at EGU 2012: Thursday 26th April, session GM2.1
13 views
Seen by: and 3 more
