Geometric Aspects of Visual Object Recognition

This thesis presents there important results in visual object recognition based on shape. (1) A new algorithm (RAST; Recognition by Adaptive Sudivisions of Tranformation space) is presented that has lower average-case complexity than any known recognition algorithm. (2) It is shown, both theoretically and empirically, that representing 3D objects as collections of 2D views (the "View-Based Approximation") is feasible and affects the reliability of 3D recognition systems no more than other commonly made approximations. (3) The problem of recognition in cluttered scenes is considered from a Bayesian perspective; the commonly-used "bounded-error errorsmeasure" is demonstrated to correspond to an independence assumption. It is shown that by modeling the statistical properties of real-scenes better, objects can be recognized more reliably.

The Individual is Nothing, the Class Everything: Psychophysics and Modeling of Recognition in Obect Classes

Most psychophysical studies of object recognition have focussed on the recognition and representation of individual objects subjects had previously explicitely been trained on. Correspondingly, modeling studies have often employed a 'grandmother'-type representation where the objects to be recognized were represented by individual units. However, objects in the natural world are commonly members of a class containing a number of visually similar objects, such as faces, for which physiology studies have provided support for a representation based on a sparse population code, which permits generalization from the learned exemplars to novel objects of that class. In this paper, we present results from psychophysical and modeling studies intended to investigate object recognition in natural ('continuous') object classes. In two experiments, subjects were trained to perform subordinate level discrimination in a continuous object class - images of computer-rendered cars - created using a 3D morphing system. By comparing the recognition performance of trained and untrained subjects we could estimate the effects of viewpoint-specific training and infer properties of the object class-specific representation learned as a result of training. We then compared the experimental findings to simulations, building on our recently presented HMAX model of object recognition in cortex, to investigate the computational properties of a population-based object class representation as outlined above. We find experimental evidence, supported by modeling results, that training builds a viewpoint- and class-specific representation that supplements a pre-existing repre-sentation with lower shape discriminability but possibly greater viewpoint invariance.

Stimulus Simplification and Object Representation: A Modeling Study

Tsunoda et al. (2001) recently studied the nature of object representation in monkey inferotemporal cortex using a combination of optical imaging and extracellular recordings. In particular, they examined IT neuron responses to complex natural objects and "simplified" versions thereof. In that study, in 42% of the cases, optical imaging revealed a decrease in the number of activation patches in IT as stimuli were "simplified". However, in 58% of the cases, "simplification" of the stimuli actually led to the appearance of additional activation patches in IT. Based on these results, the authors propose a scheme in which an object is represented by combinations of active and inactive columns coding for individual features. We examine the patterns of activation caused by the same stimuli as used by Tsunoda et al. in our model of object recognition in cortex (Riesenhuber 99). We find that object-tuned units can show a pattern of appearance and disappearance of features identical to the experiment. Thus, the data of Tsunoda et al. appear to be in quantitative agreement with a simple object-based representation in which an object's identity is coded by its similarities to reference objects. Moreover, the agreement of simulations and experiment suggests that the simplification procedure used by Tsunoda (2001) is not necessarily an accurate method to determine neuronal tuning.

Component based recognition of objects in an office environment

We present a component-based approach for recognizing objects under large pose changes. From a set of training images of a given object we extract a large number of components which are clustered based on the similarity of their image features and their locations within the object image. The cluster centers build an initial set of component templates from which we select a subset for the final recognizer. In experiments we evaluate different sizes and types of components and three standard techniques for component selection. The component classifiers are finally compared to global classifiers on a database of four objects.

Learning Three-Dimensional Shape Models for Sketch Recognition

Artifacts made by humans, such as items of furniture and houses, exhibit an enormous amount of variability in shape. In this paper, we concentrate on models of the shapes of objects that are made up of fixed collections of sub-parts whose dimensions and spatial arrangement exhibit variation. Our goals are: to learn these models from data and to use them for recognition. Our emphasis is on learning and recognition from three-dimensional data, to test the basic shape-modeling methodology. In this paper we also demonstrate how to use models learned in three dimensions for recognition of two-dimensional sketches of objects.

Teaching an Old Robot New Tricks: Learning Novel Tasks via Interaction with People and Things

As AI has begun to reach out beyond its symbolic, objectivist roots into the embodied, experientialist realm, many projects are exploring different aspects of creating machines which interact with and respond to the world as humans do. Techniques for visual processing, object recognition, emotional response, gesture production and recognition, etc., are necessary components of a complete humanoid robot. However, most projects invariably concentrate on developing a few of these individual components, neglecting the issue of how all of these pieces would eventually fit together. The focus of the work in this dissertation is on creating a framework into which such specific competencies can be embedded, in a way that they can interact with each other and build layers of new functionality. To be of any practical value, such a framework must satisfy the real-world constraints of functioning in real-time with noisy sensors and actuators. The humanoid robot Cog provides an unapologetically adequate platform from which to take on such a challenge. This work makes three contributions to embodied AI. First, it offers a general-purpose architecture for developing behavior-based systems distributed over networks of PC's. Second, it provides a motor-control system that simulates several biological features which impact the development of motor behavior. Third, it develops a framework for a system which enables a robot to learn new behaviors via interacting with itself and the outside world. A few basic functional modules are built into this framework, enough to demonstrate the robot learning some very simple behaviors taught by a human trainer. A primary motivation for this project is the notion that it is practically impossible to build an "intelligent" machine unless it is designed partly to build itself. This work is a proof-of-concept of such an approach to integrating multiple perceptual and motor systems into a complete learning agent.

Model-Based Matching by Linear Combinations of Prototypes

We describe a method for modeling object classes (such as faces) using 2D example images and an algorithm for matching a model to a novel image. The object class models are "learned'' from example images that we call prototypes. In addition to the images, the pixelwise correspondences between a reference prototype and each of the other prototypes must also be provided. Thus a model consists of a linear combination of prototypical shapes and textures. A stochastic gradient descent algorithm is used to match a model to a novel image by minimizing the error between the model and the novel image. Example models are shown as well as example matches to novel images. The robustness of the matching algorithm is also evaluated. The technique can be used for a number of applications including the computation of correspondence between novel images of a certain known class, object recognition, image synthesis and image compression.

Example Based Learning for View-Based Human Face Detection

We present an example-based learning approach for locating vertical frontal views of human faces in complex scenes. The technique models the distribution of human face patterns by means of a few view-based "face'' and "non-face'' prototype clusters. At each image location, the local pattern is matched against the distribution-based model, and a trained classifier determines, based on the local difference measurements, whether or not a human face exists at the current image location. We provide an analysis that helps identify the critical components of our system.

How are Three-Deminsional Objects Represented in the Brain?

We discuss a variety of object recognition experiments in which human subjects were presented with realistically rendered images of computer-generated three-dimensional objects, with tight control over stimulus shape, surface properties, illumination, and viewpoint, as well as subjects' prior exposure to the stimulus objects. In all experiments recognition performance was: (1) consistently viewpoint dependent; (2) only partially aided by binocular stereo and other depth information, (3) specific to viewpoints that were familiar; (4) systematically disrupted by rotation in depth more than by deforming the two-dimensional images of the stimuli. These results are consistent with recently advanced computational theories of recognition based on view interpolation.

Biologically Plausible Neural Circuits for Realization of Maximum Operations

Object recognition in the visual cortex is based on a hierarchical architecture, in which specialized brain regions along the ventral pathway extract object features of increasing levels of complexity, accompanied by greater invariance in stimulus size, position, and orientation. Recent theoretical studies postulate a non-linear pooling function, such as the maximum (MAX) operation could be fundamental in achieving such invariance. In this paper, we are concerned with neurally plausible mechanisms that may be involved in realizing the MAX operation. Four canonical circuits are proposed, each based on neural mechanisms that have been previously discussed in the context of cortical processing. Through simulations and mathematical analysis, we examine the relative performance and robustness of these mechanisms. We derive experimentally verifiable predictions for each circuit and discuss their respective physiological considerations.

Generalization over contrast and mirror reversal, but not figure-ground reversal, in an "edge-based

Baylis & Driver (Nature Neuroscience, 2001) have recently presented data on the response of neurons in macaque inferotemporal cortex (IT) to various stimulus transformations. They report that neurons can generalize over contrast and mirror reversal, but not over figure-ground reversal. This finding is taken to demonstrate that ``the selectivity of IT neurons is not determined simply by the distinctive contours in a display, contrary to simple edge-based models of shape recognition'', citing our recently presented model of object recognition in cortex (Riesenhuber & Poggio, Nature Neuroscience, 1999). In this memo, I show that the main effects of the experiment can be obtained by performing the appropriate simulations in our simple feedforward model. This suggests for IT cell tuning that the possible contributions of explicit edge assignment processes postulated in (Baylis & Driver, 2001) might be smaller than expected.

Global Depth Perception from Familiar Scene Structure

In the absence of cues for absolute depth measurements as binocular disparity, motion, or defocus, the absolute distance between the observer and a scene cannot be measured. The interpretation of shading, edges and junctions may provide a 3D model of the scene but it will not inform about the actual "size" of the space. One possible source of information for absolute depth estimation is the image size of known objects. However, this is computationally complex due to the difficulty of the object recognition process. Here we propose a source of information for absolute depth estimation that does not rely on specific objects: we introduce a procedure for absolute depth estimation based on the recognition of the whole scene. The shape of the space of the scene and the structures present in the scene are strongly related to the scale of observation. We demonstrate that, by recognizing the properties of the structures present in the image, we can infer the scale of the scene, and therefore its absolute mean depth. We illustrate the interest in computing the mean depth of the scene with application to scene recognition and object detection.

Categorization in IT and PFC: Model and Experiments

In a recent experiment, Freedman et al. recorded from inferotemporal (IT) and prefrontal cortices (PFC) of monkeys performing a "cat/dog" categorization task (Freedman 2001 and Freedman, Riesenhuber, Poggio, Miller 2001). In this paper we analyze the tuning properties of view-tuned units in our HMAX model of object recognition in cortex (Riesenhuber 1999) using the same paradigm and stimuli as in the experiment. We then compare the simulation results to the monkey inferotemporal neuron population data. We find that view-tuned model IT units that were trained without any explicit category information can show category-related tuning as observed in the experiment. This suggests that the tuning properties of experimental IT neurons might primarily be shaped by bottom-up stimulus-space statistics, with little influence of top-down task-specific information. The population of experimental PFC neurons, on the other hand, shows tuning properties that cannot be explained just by stimulus tuning. These analyses are compatible with a model of object recognition in cortex (Riesenhuber 2000) in which a population of shape-tuned neurons provides a general basis for neurons tuned to different recognition tasks.

Investigating shape representation in area V4 with HMAX: Orientation and Grating selectivities

The question of how shape is represented is of central interest to understanding visual processing in cortex. While tuning properties of the cells in early part of the ventral visual stream, thought to be responsible for object recognition in the primate, are comparatively well understood, several different theories have been proposed regarding tuning in higher visual areas, such as V4. We used the model of object recognition in cortex presented by Riesenhuber and Poggio (1999), where more complex shape tuning in higher layers is the result of combining afferent inputs tuned to simpler features, and compared the tuning properties of model units in intermediate layers to those of V4 neurons from the literature. In particular, we investigated the issue of shape representation in visual area V1 and V4 using oriented bars and various types of gratings (polar, hyperbolic, and Cartesian), as used in several physiology experiments. Our computational model was able to reproduce several physiological findings, such as the broadening distribution of the orientation bandwidths and the emergence of a bias toward non-Cartesian stimuli. Interestingly, the simulation results suggest that some V4 neurons receive input from afferents with spatially separated receptive fields, leading to experimentally testable predictions. However, the simulations also show that the stimulus set of Cartesian and non-Cartesian gratings is not sufficiently complex to probe shape tuning in higher areas, necessitating the use of more complex stimulus sets.

Face processing in humans is compatible with a simple shape-based model of vision

Understanding how the human visual system recognizes objects is one of the key challenges in neuroscience. Inspired by a large body of physiological evidence (Felleman and Van Essen, 1991; Hubel and Wiesel, 1962; Livingstone and Hubel, 1988; Tso et al., 2001; Zeki, 1993), a general class of recognition models has emerged which is based on a hierarchical organization of visual processing, with succeeding stages being sensitive to image features of increasing complexity (Hummel and Biederman, 1992; Riesenhuber and Poggio, 1999; Selfridge, 1959). However, these models appear to be incompatible with some well-known psychophysical results. Prominent among these are experiments investigating recognition impairments caused by vertical inversion of images, especially those of faces. It has been reported that faces that differ "featurally" are much easier to distinguish when inverted than those that differ "configurally" (Freire et al., 2000; Le Grand et al., 2001; Mondloch et al., 2002) ??finding that is difficult to reconcile with the aforementioned models. Here we show that after controlling for subjects' expectations, there is no difference between "featurally" and "configurally" transformed faces in terms of inversion effect. This result reinforces the plausibility of simple hierarchical models of object representation and recognition in cortex.