Object Recognition (Pattern Recognition)

Top-down facilitation of visual object recognition: object-based and context-based contributions

by Mark Fenske

Topographic feature mapping for head pose estimation with application to facial gesture interfaces

by Bisser Raytchev

in Lecture Notes in Computer Science, HCI/ICCV, LNCS 3766, pp. 180-188, Springer-Verlag, 2005.

We propose a new general approach to the problem of head pose estimation,
based on semi-supervised... more We propose a new general approach to the problem of head pose estimation,
based on semi-supervised low-dimensional topographic feature mapping.
We show how several recently proposed nonlinear manifold learning
methods can be applied in this general framework, and additionally, we present
a new algorithm, IsoScale, which combines the best aspects of some of the other
methods. The efficacy of the proposed approach is illustrated both on a viewand
illumination-varied face database, and in a real-world human-computer interface
application, as head pose based facial-gesture interface for automatic
wheelchair navigation.

View-Invariant Object Recognition with Visibility Maps

by Bisser Raytchev

in Proc. 20-th International Conference on Pattern Recognition ICPR’2010, pp. 1040-1043, Istanbul, Turkey, 2010.

In this paper we propose a new framework for view-invariant 3D object recognition, based on what we call Visibility... more In this paper we propose a new framework for view-invariant 3D object recognition, based on what we call Visibility Maps. A Visibility Map (VM) encodes a compact model of an arbitrary 3D object for which a set of images taken from different views is available. Representative local invariant features extracted from each image are selectively combined to form a visibility basis, in terms of which an arbitrary view of the modeled object can be represented. A metric which incorporates geometric information is also provided for comparing test images to the model, and can be used for recognition.

Class-specific low-dimensional representation of local features for viewpoint invariant object recognition

by Bisser Raytchev

in Proc. 10-th Asian Conference on Computer Vision ACCV’2010, Queenstown, New Zealand, vol. 3, pp. 1552-1563, 2010.

In this paper we propose a new general framework to obtain
more distinctive local invariant features by... more In this paper we propose a new general framework to obtain
more distinctive local invariant features by projecting the original
feature descriptors into low–dimensional feature space, while simultaneously
incorporating also class information. In the resulting feature space,
the features from different objects project to separate areas, while locally
the metric relations between features corresponding to the same object
are preserved. The low–dimensional feature embedding is obtained by
a modified version of classical Multidimensional Scaling, which we call
supervised Multidimensional Scaling (sMDS). Experimental results on a
database containing images of several different objects with large variation
in scale, viewpoint, illumination conditions and background clutter
support the view that embedding class information into the feature representation
is beneficial and results in more accurate object recognition.

Spatial Query and Object Recognition to Support 3D Digital Building Model Use

by Andre Chaszar

prepared for and presented at the18th EG-ICE International Workshop

Intelligent Computing in Engineering (ICE11)
University of Twente, The Netherlands
6th to 8th July 2011

http://www.utwente.nl/ctw/eg-ice/

In AEC practice as well as in academic settings the use of 3-dimensional digital building models is becoming... more In AEC practice as well as in academic settings the use of 3-dimensional digital building models is becoming increasingly prevalent, and the models used tend to be increasingly large and complex. While the use of such models is to some extent enabling collaborative processes of design, analysis, communication and decision-making which were not feasible previously - when analog and 2D digital representations were the norm - it has been widely observed that the use of these 3D models also poses some significant difficulties, particularly regarding the exchange of models in multi-party, multi-disciplinary design processes. The types of and reasons for these difficulties are numerous and varied, but the one of specific interest here concerns the ways in which the data contained in these models is organized and retrieved. The present paper addresses these two interlinked issues and proposes some improvements to the current state of affairs by way of geometric-content-based methods for analyzing and querying the model objects, which can complement the existing text- and other assigned-attribute-based methods commonly employed. Some basic considerations regarding geometric-content-based search and classification are presented, followed by description of tests carried out to assess the implementation of some proposed methods and discussion of the test results' implications for further work in this area.

Sexual orientation and spatial position effects on selective forms of object location memory.

by Qazi Rahman

Brain Cogn. 2011 Apr;75(3):217-24.

Seeing different objects in different ways: Measuring ventral visual tuning to sensory and semantic features with dynamically adaptive imaging

by Michele Veldsman

'Breaking'position-invariant object recognition

by David Cox

A set of high quality colour images with Spanish norms for seven relevant psycholinguistic variables: The Nombela naming test

by Keith Laws

This paper presents a new corpus of 140 high quality colour images belonging to 14 subcategories and covering a range... more This paper presents a new corpus of 140 high quality colour images belonging to 14 subcategories and covering a range of naming difficulty. One hundred and six Spanish speakers named the items and provided data for several psycholinguistic variables: age of acquisition, familiarity, manipulability, name agreement, typicality and visual complexity.
Furthermore, we also present lexical frequency data derived internet search hits. Apart from the large number of variables evaluated, these stimuli present an important advantage with respect to other comparable image corpora in so far as naming performance in healthy individuals is less prone to ceiling effect problems. Reliability and validity indexes showed that our items display similar psycholinguistic characteristics to those of other corpora. In sum, this set of ecologically valid stimuli provides a useful tool for scientists engaged in cognitive and neuroscience-based research.

Naming Without Knowing and Appearance Without Associations: Evidence for Constructive Processes In Semantic Memory?

by Keith Laws

This study describes a patient (SE) with temporal lobe injury resulting from Herpes Simplex Encephalitis, who... more This study describes a patient (SE) with temporal lobe injury resulting from Herpes Simplex Encephalitis, who displayed a previously unreported impairment in which his knowledge of associative and functional attributes of animals was disproportionately impaired by comparison with his knowledge of their sensory attributes (including their visual properties and characteristic sounds). His knowledge of man-made objects was preserved. A striking aspect of the present case was that the patient remained able to name many animals from their pictures, despite making gross errors in generating associative information about these same animals. This suggests that a semantic representation incorporating stored sensory knowledge may be sufficient for naming (at least for biological categories) and associative information may be unnecessary. Semantic knowledge may normally incorporate more information than is necessary for identification. SE's errors were found to be confabulatory and reconstructive in nature and it is argued that this aspect of his performance challenges passive conceptions of semantic memory couched in terms of a catalogue of stored representations. It is proposed that the patient's disorder affects a dynamic, constructive, and inferential component of his knowledge base, and that this component is sensitive to semantic category.

Visual Object Perception In Schizophrenia: Further Evidence for a Selective Impairment In Semantic Memory

by Keith Laws

Early studies suggested visual form perception impairment in schizophrenia. To re-examine this claim and characterise... more Early studies suggested visual form perception impairment in schizophrenia. To re-examine this claim and characterise the deficit neuropsychologically, 41 schizophrenic patients were administered tests sensitive to different levels of visual object perception and recognition. Intellectually well-preserved patients were examined separately on these and additional tests. Single case analysis was also applied to four patients showing varying degrees of general intellectual impairment. As a group, the patients showed little impairment on tests of early visual object perception, but greater impairment on higher-level visual processing tests, in particular object recognition and naming. This held true for patients with preserved general intellectual function. Single case analysis suggested that patients with schizophrenia have a selective deficit affecting object recognition and identification, with a pattern similar to visual associative agnosia in neurological patients

Loss of Stored Knowledge of Object Structure: Implications for" Category-Specific" Deficits

by Keith Laws

Turnbull OH & Laws KR (2000)

Following a right-hemisphere lesion, the patient SM had impaired object recognition, with good elementary visual... more Following a right-hemisphere lesion, the patient SM had impaired object recognition, with good elementary visual abilities, and could derive information about object structure. He was also impaired on all tasks tapping stored structural knowledge, even when tested in the verbal modality. This suggests that SM has a disorder affecting stored knowledge of object structure, though he remains able to assemble novel structural descriptions. His object recognition ability also appeared significantly worse for non-living things. By contrast, existing models relating to stored knowledge would predict that SM would show greater impairment with living things. We argue that SM's deficit reflects the loss of a type of structural knowledge that relates to the “within-item structural diversity” of items. It is argued that living things show less structural variation than objects in the natural world, and might arguably be easier to recognise, because the image of the to-be-recognised object would be similar to the stored representation. Hence, a deficit affecting this aspect of stored knowledge would differentially impact upon non-living things. This argument receives confirming independent support from the finding that normal subjects ratings for the within-item structural diversity of visual stimuli are (unlike other “critical” variables) significant predictors of SM's naming performance

Object Recognition Without Knowledge of Object Orientation

by Keith Laws

Several theories have been proposed to explain our ability to recognise objects from a number of viewpoints.... more Several theories have been proposed to explain our ability to recognise objects from a number of viewpoints. Orientation-dependent accounts emphasize the position of the object relative to the viewer, while orientation-independent accounts (e.g. Marr) rely on descriptions of an object's component parts relative to its principal axis of elongation. An opportunity to compare the merit of these theories has arisen in a patient (L.G.) who had a rare neuropsychological sign in which knowledge of the canonical upright of object drawings was profoundly disrupted. Such orientation errors were evident in her drawings from memory and to copy, and in an orientation-matching task. In a critical experiment she showed a deficit in providing the canonical upright of individual object drawings that was independent of any difficulty in object recognition. The implications of these data for theories of object recognition are discussed

Visual Object Processing In Schizophrenia: Evidence for An Associative Agnosic Deficit

by Keith Laws

Early studies suggested visual form perception impairment in schizophrenia. To re-examine this claim and characterise... more Early studies suggested visual form perception impairment in schizophrenia. To re-examine this claim and characterise the deficit neuropsychologically, 41 schizophrenic patients were administered tests sensitive to different levels of visual object perception and recognition. Intellectually well-preserved patients were examined separately on these and additional tests. Single case analysis was also applied to four patients showing varying degrees of general intellectual impairment. As a group, the patients showed little impairment on tests of early visual object perception, but greater impairment on higher-level visual processing tests, in particular object recognition and naming. This held true for patients with preserved general intellectual function. Single case analysis suggested that patients with schizophrenia have a selective deficit affecting object recognition and identification, with a pattern similar to visual associative agnosia in neurological patients

The Effect of “masking” on Picture Naming

by Keith Laws

It is frequently assumed that because compared to nonliving things, living things are less familiar, have lower name... more It is frequently assumed that because compared to nonliving things, living things are less familiar, have lower name frequency, and are more visually complex, this makes them
more difficult to name by patients and normal subjects. This has also been implicitly accepted as an explanation for the greater incidence of living thing disorders. Patient studies
do not, however, typically contain any premorbid data and so, we do not know that the same variables would have necessarily predicted their ‘normal’ performance. To examine
this issue, we measured picture-naming latencies in normal subjects presented with unmasked and masked versions of the same line drawings. In accord with other recent studies, living things were named faster than nonliving things. Furthermore, contrary to some theories of category naming, the living thing advantage persisted regardless of whether stimuli were undegraded, degraded or the density of degradation. Finally, multiple simultaneous regression analyses showed that one visual variable (Euclidean Overlap) and one linguistic variable (Age of Acquisition) predicted naming latencies across all masked and unmasked conditions. Other variables either had no predictive value (Contour Overlap; Name Frequency; Category); predicted only high masking (Visual Complexity; Familiarity), or normal and low masking (Number of Phonemes). These findings imply that the more commonly documented deficits for living things do not reflect an exaggeration of the normal profile (be it with masked or unmasked stimuli) or the influence of the same variables that
affect normal naming.

Visual processing in Alzheimer’s disease: Surface detailand colour fail to aid object identification

by Keith Laws

It has been suggested that object recognition in patients with Alzheimer’s disease (AD) may be strongly influenced... more It has been suggested that object recognition in patients with Alzheimer’s disease (AD) may be strongly influenced both by image format (e.g. colour vs. line-drawn) and by low-level visual impairments. To examine these notions, we tested basic visual functioning and picture naming in 41 AD patients and 40
healthy elderly controls. Picture naming was examined using 105 images representing a wide range of living and nonliving subcategories (from the Hatfield image test [HIT]: [Adlington, R. A., Laws, K. R., & Gale, T. M. (2009). The Hatfield image test (HIT): A new picture test and norms for experimental and clinical use. Journal of Clinical and Experimental Neuropsychology]), with each itempresented in colour, greyscale, or line-drawn formats. Whilst naming for elderly controls improved linearly with the addition of surface detail and colour, ADpatients showed no benefit fromthe addition of either surface information or colour.Additionally, controls showed a significant category by format interaction; however, the same profile did not emerge for AD patients. Finally, AD patients showed widespread and significant impairment on tasks of visual functioning, and low-level visual impairment was predictive of patient naming

Gender Affects Naming Latencies for Living and Nonliving Things: Implications for Familiarity.

by Keith Laws

Laws (1999)

Recent studies indicate the presence of a gender-by-category interaction in the naming abilities of both Alzheimer’s... more Recent studies indicate the presence of a gender-by-category interaction in the naming abilities of both Alzheimer’s patients and normal subjects (Laiacona, Barbarotto and Capitani, 1998; McKenna and Parry, 1994). In particular, males appear to be better than females at naming nonliving things and females better at naming living things. Similarly, in a recent study of semantic fluency, males retrieved more names of tools than females and females more names of fruit than males (Capitani, Laiacona and Barbarotto, 1999). Such findings have important implications for our understanding of category-specific disorders. The current study examined the naming latencies of normal subjects to pictures of living and nonliving things. We confirm a gender-by category interaction across both subject and item, with females being slower than males to name nonliving things and males slower to name living things. This finding could not be explained by differential difficulty of items or differences in gender-based familiarity ratings.

Crowded and Sparse Domains In Object Recognition: Consequences for Categorization and Naming

by Keith Laws

Some models of object recognition propose that items from structurally crowded categories (e.g., living things) permit... more Some models of object recognition propose that items from structurally crowded categories (e.g., living things) permit faster access to superordinate semantic information than structurally dissimilar categories (e.g., nonliving things), but slower access to individual object information when naming items. We present four experiments that utilize the same matched stimuli: two examine superordinate categorization and two examine picture naming. Experiments 1 and 2 required participants to sort pictures into their appropriate superordinate categories and both revealed faster categorization for living than nonliving things. Nonetheless, the living thing superiority disappeared when the atypical categories of body parts and musical instruments were excluded. Experiment 3 examined naming latency and found no difference between living and nonliving things. This finding was replicated in Experiment 4 where the same items were presented in different formats (e.g., color and line-drawn versions). Taken as a whole, these experiments show that the ease with which people categorize items maps strongly onto the ease with which they name them.

Basic-Level Visual Similarity and Category Specificity

by Keith Laws

Category-Specificity Can Emerge From Bottom-Up Visual Characteristics: Evidence From a Modular Neural Network

by Keith Laws

The role of bottom-up visual processes in category-specific object recognition has been largely unexplored. We... more The role of bottom-up visual processes in category-specific object recognition has been largely unexplored. We examined the role of low-level visual characteristics in category specific recognition using a modular neural network comprising both unsupervised and supervised components. One hundred standardised pictures from ten different categories (Five living and five nonliving, including body parts and musical instruments) were presented to a Kohonen self-organising map (SOM) which re-represents the visual stimuli by clustering them within a smaller number of dimensions. The SOM representations were then used to train an attractor network to learn the superordinate category of each item. The ease with which the model acquired the category mappings was investigated with respect to emerging category effects. We found that the superordinates could be separated by very low-level visual factors (as extracted by the SOM). The model also accounted for the well documented atypicality of body parts and musical instrument superordinates. The model has clear relevance to human object recognition since the model was quicker to learn more typical category exemplars and finally the model also accounted for more than 20% of the naming variance in a sample of 57 brain injured subjects. We conclude that purely bottom-up visual characteristics can explain some important features of category-specific phenomena.