Learning and Exploiting Camera Geometry for Computer Vision

This work explores the use of statistical methods in describing and estimating camera poses, as well as the information feedback loop between camera pose and object detection. Surging development in robotics and computer vision has pushed the need for algorithms that infer, understand, and utilize information about the position and orientation of the sensor platforms when observing and/or interacting with their environment.

The first contribution of this thesis is the development of a set of statistical tools for representing and estimating the uncertainty in object poses. A distribution for representing the joint uncertainty over multiple object positions and orientations is described, called the mirrored normal-Bingham distribution. This distribution generalizes both the normal distribution in Euclidean space, and the Bingham distribution on the unit hypersphere. It is shown to inherit many of the convenient properties of these special cases: it is the maximum-entropy distribution with fixed second moment, and there is a generalized Laplace approximation whose result is the mirrored normal-Bingham distribution. This distribution and approximation method are demonstrated by deriving the analytical approximation to the wrapped-normal distribution. Further, it is shown how these tools can be used to represent the uncertainty in the result of a bundle adjustment problem.

Another application of these methods is illustrated as part of a novel camera pose estimation algorithm based on object detections. The autocalibration task is formulated as a bundle adjustment problem using prior distributions over the 3D points to enforce the objects' structure and their relationship with the scene geometry. This framework is very flexible and enables the use of off-the-shelf computational tools to solve specialized autocalibration problems. Its performance is evaluated using a pedestrian detector to provide head and foot location observations, and it proves much faster and potentially more accurate than existing methods.

Finally, the information feedback loop between object detection and camera pose estimation is closed by utilizing camera pose information to improve object detection in scenarios with significant perspective warping. Methods are presented that allow the inverse perspective mapping traditionally applied to images to be applied instead to features computed from those images. For the special case of HOG-like features, which are used by many modern object detection systems, these methods are shown to provide substantial performance benefits over unadapted detectors while achieving real-time frame rates, orders of magnitude faster than comparable image warping methods.

The statistical tools and algorithms presented here are especially promising for mobile cameras, providing the ability to autocalibrate and adapt to the camera pose in real time. In addition, these methods have wide-ranging potential applications in diverse areas of computer vision, robotics, and imaging.

Other Than a Citizen: Vernacular Poetics in Postwar America

Few symbols of 1950s-1960s America remain as central to our contemporary conception of Cold War culture as the iconic ranch-style suburban home. While the house took center stage in the Nixon/Khrushchev kitchen debates as a symbol of modern efficiency and capitalist values, its popularity depended largely upon its obvious appropriation of vernacular architecture from the 19th century, those California haciendas and Texas dogtrots that dotted the American west. Contractors like William Levitt modernized the historical common houses, hermetically sealing their porous construction, all while using the ranch-style roots of the dwelling to galvanize a myth of an indigenous American culture. At a moment of intense occupational bureaucracy, political uncertainty and atomized social life, the rancher gave a self-identifying white consumer base reason to believe they could master their own plot in the expansive frontier. Only one example of America’s mid-century love affair with commodified vernacular forms, the ranch-style home represents a broad effort on the part of corporate and governmental interest groups to transform the vernacular into a style that expresses a distinctly homogenous vision of American culture. “Other than a Citizen” begins with an anatomy of that transformation, and then turns to the work of four poets who sought to reclaim the vernacular from that process of standardization and use it to countermand the containment-era strategies of Cold War America.

In four chapters, I trace references to common speech and verbal expressivity in the poetry and poetic theory of Charles Olson, Robert Duncan, LeRoi Jones/Amiri Baraka and Gwendolyn Brooks, against the historical backdrop of the Free-Speech Movement and the rise of mass-culture. When poets frame nonliterary speech within the literary page, they encounter the inability of writing to capture the vital ephemerality of verbal expression. Rather than treat this limitation as an impediment, the writers in my study use the poem to dramatize the fugitivity of speech, emphasizing it as a disruptive counterpoint to the technologies of capture. Where critics such as Houston Baker interpret the vernacular strictly in terms of resistance, I take a cue from the poets and argue that the vernacular, rooted etymologically at the intersection of domestic security and enslaved margin, represents a gestalt form, capable at once of establishing centralized power and sparking minor protest. My argument also expands upon Michael North’s exploration of the influence of minstrelsy and regionalism on the development of modernist literary technique in The Dialect of Modernism. As he focuses on writers from the early 20th century, I account for the next generation, whose America was not a culturally inferior collection of immigrants but an imperial power, replete with economic, political and artistic dominance. Instead of settling for an essentially American idiom, the poets in my study saw in the vernacular not phonetic misspellings, slang terminology and fragmented syntax, but the potential to provoke and thereby frame a more ethical mode of social life, straining against the regimentation of citizenship.

My attention to the vernacular argues for an alignment among writers who have been segregated by the assumption that race and aesthetics are mutually exclusive categories. In reading these writers alongside one another, “Other than a Citizen” shows how the avant-garde concepts of projective poetics and composition by field develop out of an interest in black expressivity. Conversely, I trace black radicalism and its emphasis on sociality back to the communalism practiced at the experimental arts college in Black Mountain, North Carolina, where Olson and Duncan taught. In pressing for this connection, my work reveals the racial politics embedded within the speech-based aesthetics of the postwar era, while foregrounding the aesthetic dimension of militant protest.

Not unlike today, the popular rhetoric of the Cold War insists that to be a citizen involves defending one’s status as a rightful member of an exclusionary nation. To be other than a citizen, as the poets in my study make clear, begins with eschewing the false certainty that accompanies categorical nominalization. In promoting a model of mutually dependent participation, these poets lay the groundwork for an alternative model of civic belonging, where volition and reciprocity replace compliance and self-sufficiency. In reading their lines, we become all the more aware of the cracks that run the length of our load-bearing walls.