Showing papers by "Andrew C. Gallagher published in 2014"

GPS Refinement and Camera Orientation Estimation from a Single Image and a 2D Map

Abstract: A framework is presented for refining GPS location and estimate the camera orientation using a single urban building image, a 2D city map with building outlines, given a noisy GPS location. We propose to use tilt-invariant vertical building corner edges extracted from the building image. A location-orientation hypothesis, which we call an LOH, is a proposed map location from which an image of building corners would occur at the observed positions of corner edges in the photo. The noisy GPS location is refined and orientation is estimated using the computed LOHs. Experiments show the framework improves GPS accuracy significantly, generally produces reliable orientation estimation, and is computationally efficient.

The Hidden Sides of Names—Face Modeling with First Name Attributes

Abstract: This paper introduces the new idea of describing people using first names. We show that describing people in terms of similarity to a vector of possible first names is a powerful representation of facial appearance that can be used for a number of important applications, such as naming never-seen faces and building facial attribute classifiers. We build models for 100 common first names used in the US and for each pair, construct a pairwise first-name classifier. These classifiers are built using training images downloaded from the internet, with no additional user interaction. This gives our approach important advantages in building practical systems that do not require additional human intervention for data labeling. The classification scores from each pairwise name classifier can be used as a set of facial attributes to describe facial appearance. We show several surprising results. Our name attributes predict the correct first names of test faces at rates far greater than chance. The name attributes are applied to gender recognition and to age classification, outperforming state-of-the-art methods with all training images automatically gathered from the internet. We also demonstrate the powerful use of our name attributes for associating faces in images with names from caption, and the important application of unconstrained face verification.

Discovering visited travel destinations from a set of digital images

Abstract: Aspects of the disclosure relate to identifying visited travel destinations from a set of digital images associated with users of a social networking system. For example, one or more computing devices provide access to an individual user's account, including the individual user and other users affiliated with the individual user via the social networking system. One or more digital images are received from a computing device associated with the individual user and from one or more second computing devices associated with the other users of the social networking system. From each digital image, a geo-location is determined for each digital image. The one or more computing devices display each geo-located image on a map at a position corresponding to the determined geo-location for the geo-located image.

Putting the User in the Loop for Image-Based Modeling

Abstract: We refer to the task of recovering the 3D structure of an object or a scene using 2D images as image-based modeling. In this paper, we formulate the task of recovering the 3D structure as a discrete optimization problem solved via energy minimization. In this standard framework of a Markov random field (MRF) defined over the image we present algorithms that allow the user to intuitively interact with the algorithm. We introduce an algorithm where the user guides the process of image-based modeling to find and model the object of interest by manually interacting with the nodes of the graph. We develop end user applications using this algorithm that allow object of interest 3D modeling on a mobile device and 3D printing of the object of interest. We also propose an alternate active learning algorithm that guides the user input. An initial attempt is made at reconstructing the scene without supervision. Given the reconstruction, an active learning algorithm uses intuitive cues to quantify the uncertainty of the algorithm and suggest regions, querying the user to provide support for the uncertain regions via simple scribbles. These constraints are used to update the unary and the pairwise energies that, when solved, lead to better reconstructions. We show through machine experiments and a user study that the proposed approach intelligently queries the users for constraints, and users achieve better reconstructions of the scene faster, especially for scenes with textureless surfaces lacking strong textural or structural cues that algorithms typically require.

Determining Compass Orientation Of Imagery

Abstract: The orientation of imagery relative to a compass bearing may be determined based on the position of the sun or other information relating to celestial bodies captured in the image

Methods and systems for annotating images based on special events

Abstract: A method for associating event times or time periods with digital images in a collection for determining if a digital image is of interest, includes storing a collection of digital images each having an associated capture time; comparing the associated capture time in the collection with a special event time to determine if a digital image in the collection is of interest, wherein the comparing step includes calculation of a special event time associated with a special event based on the calendar time associated with the special event and using such information to perform the comparison step; and associating digital images of interest with the special event.

Recognizing People in Social Context

Abstract: The people in an image are generally not strangers, but instead often share social relationships such as husband-wife, siblings, grandparent-child, father-child, or mother-child. Further, the social relationship between a pair of people influences the relative position and appearance of the people in the image. This chapter explores using familial social relationships as context, recognizing people and recognizing the social relationships between pairs of people. We introduce a model for representing the interaction between social relationship, facial appearance, and identity. The experiments on a set of personal collections show that significant improvement in people recognition is achieved by modeling social relationships, even in a weak label setting that is attractive in practical applications. Furthermore, we show that social relationships are effectively recognized in images from a separate test image collection.