New Submissions
Automatic Understanding of Image and Video Advertisements
Author: Adriana Kovashka
There is more to images than their objective physical content: for example, advertisements are created to persuade a viewer to take a certain action. We propose the novel problem of automatic advertisement understanding. To enable research on this problem, we create two large datasets: an image
...read more
Modality Combination and Transformation for Recognizing Non-Photorealistic Objects
Author: Christopher Thomas
Computer vision systems currently lack the ability to recognize objects in novel modalities, especially when data from the target modality is limited. We propose a method for recognizing objects in novel modalities, without labeled data in that modality (or without any data whatsoever from that
...read more
Non-Semantic Attribute Transfer
Author: Nils Murrugarra
Attribute learning is accomplished with a large amount of annotations. However, what happens if for a given attribute, we do not have many annotations? The standard approach would be to perform transfer learning or domain adaptation, where we use source models trained on other attributes, howev
...read more
Technical Report OpenSalicon: An Open Source Implementation of the Salicon Saliency Model
Author: Christopher Thomas
In this technical report, we present our publicly downloadable implementation of the SALICON saliency model. At the time of this writing, SALICON is one of the top performing saliency models on the MIT 300 fixation prediction dataset which evaluates how well an algorithm is able to predict where hum
...read more
Decoding Image Advertisements: A Dataset and Four Tasks
Authors: Adriana Kovashka, Zuha Agha
Image ads constitute a domain of images that use visual rhetoric to convey a message. Understanding the message being conveyed and the strategies that ads use to convince a viewer to perform some action (e.g. buy a product), goes beyond predicting what is present in images. It involves reasoning abo
...read more
Learning Attributes from Human Gaze
Authors: Adriana Kovashka, Nils Murrugarra-Llerena
While semantic visual attributes have been shown useful for a variety of tasks, many attributes are difficult to model computationally. Some of the reasons for this difficulty are: (1) it is not clear where in an image the attribute lives, and (2) some attributes are ambiguous and have different int
...read more
Seeing Behind the Camera: Identifying the Authorship of a Photograph
Authors: Adriana Kovashka, Christopher Thomas
We introduce the novel problem of identifying the photographer behind the photograph. To explore the feasibility of current computer vision techniques to address this problem, we created a new dataset of over 180,000 images taken by 41 well-known photographers. Using this dataset, we examined the e
...read more
Parse Tree Fragmentation Evaluation Methodology
Author: Homa B. Hashemi
Statistical parsers have a tendency to assign complete well-formed syntactic analyses to a sentence even if it is not well-formed. Consequently, a resulting parse tree is not always a good match for a sentence. The goal of this project is to address the problem of assigning reasonable syntactic anal
...read more
A Dynamic-Image Framework for Geometric Modeling, Simulation and Analysis of Human Joints
Author: Md. Abedul
We propose a dynamic-image driven, computational framework for the geometric modeling and simulation of multiarticular anatomical joints. The framework uses anatomic knowledge, computational methods, static and dynamic medical images to accurately measure bone kinematics and to model the subject-spe
...read more
Navigating a Galaxy of Observations: From Frustration to Innovation
Authors: Daniel Q. Oliphant, Brian A. Cherinka, W. Michael Wood-Vasey, Jeffrey A. Newman, Alexandros Labrinidis, Panos Chrysanthis, G. Elisabeta Marai
Over the next decade the amount of information available to the typical astronomer will grow by two orders of magnitude, thanks to programs such as Pan-STARRS (Panoramic Survey Telescope and Rapid Response System) and LSST (Large Synoptic Survey Telescope). However, we lack an easy-to-use and scalab
...read more
Real Time Ray Tracing in a Space Limited Environment
Authors: Daniel Q. Oliphant, G. Elisabeta Marai
Ray tracing as a rendering technique has many useful applications in both simulation and entertainment. A real-time ray-tracing engine is even more valuable as it allows for user interactivity, greatly increasing the number of potential uses for such a renderer. Most ray tracing techniques are emp
...read more
Optimal Parsing Strategies for Linear Context-Free Rewriting Systems
Author: Daniel Gildea
Reduction is the operation of transforming a production in a Linear Context-Free Rewriting System (LCFRS) into two simpler productions by factoring out a subset of the nonterminals on the production righthand side. Reduction lowers the rank of a production but may increase its fan-out. We show how
...read more
DyC: An Expressive Annotation-Directed Dynamic Compiler for C
Authors: Brian Grant, Markus Mock, Matthai Philipose, Craig Chambers, Susan J. Eggers
We present the design of DyC, a dynamic-compilation system for C based on run-time specialization. Directed by a few declarative user annotations that specify the variables and code on which dynamic compilation should take place, a binding-time analysis computes the set of run-time constants at
...read more