Princeton > CS Dept > PIXL > Graphics > Lunch Local Access 


The PIXL lunch meets every Monday during the semester at noon in room 402 of the Computer Science building. To get on the mailing list to receive announcements, sign up for the "pixl-talks" list at lists.cs.princeton.edu.

Upcoming Talks


No talks scheduled yet.

Previous Talks


Monday, February 05, 2018
Uncovering Perceptual Priors Using Automated Serial Reproduction Chains
Thomas Langlois

Abstract
Human memory can be understood in terms of an inferential Bayesian process, where memories are the product of noisy sensory information combined with knowledge drawn from prior experience. Expectations from prior knowledge can introduce biases in the encoding of sensory information into internal representations. In this work, we used automated serial reproduction chains to simulate the transmission of information from one observer to the next, by arranging participants on amazon mechanical turk into a large number of carefully curated transmission chains as they completed a spatial memory task. While confirming some previous findings, we demonstrate that our approach paints a much more nuanced picture of spatial memory biases, revealing that spatial memory priors are often far more intricate and complex than previously thought.


Monday, February 19, 2018
ToonCap: A Layered Deformable Model for Capturing Poses From Cartoon Characters
Xinyi Fan

Abstract
Characters in traditional artwork such as children’s books or cartoon animations are typically drawn once, in fixed poses, with little opportunity to change the characters’ appearance or re-use them in a different animation. To enable these applications one can fit a consistent parametric deformable model—a puppet —to different images of a character, thus establishing consistent segmentation, dense semantic correspondence, and deformation parameters across poses. In this work we argue that a layered deformable puppet is a natural representation for hand-drawn characters, providing an effective way to deal with the articulation, expressive deformation, and occlusion that are common to this style of artwork. Our main contribution is an automatic pipeline for fitting these models to unlabeled images depicting the same character in various poses. We demonstrate that the output of our pipeline can be used directly for editing and re-targeting animations.


Monday, March 12, 2018
Animating Static Pictures with Kinetic Textures
Nora Willett

Abstract
We present an interactive tool to animate the visual elements of a static picture, based on simple sketch-based markup. While animated images enhance websites, infographics, logos, e-books, and social media, creating such animations from still pictures is difficult for novices and tedious for experts. Creating automatic tools is challenging due to ambiguities in object segmentation, relative depth ordering, and non-existent temporal information. With a few user drawn scribbles as input, our mixed initiative creative interface extracts repetitive texture elements in an image, and supports animating them. Our system also facilitates creation of multiple layers to enhance depth cues in the animation. Finally, after analyzing the artwork during segmentation, several animation processes automatically generate kinetic textures cite{Kazi:2014:DBL:2556288.2556987} that are spatio-temporally coherent with the source image. Our results as well as feedback from our user evaluation sugg est that our system effectively allows illustrators and animators to add life to still images in a broad range of visual styles.


Monday, April 02, 2018
LandmarkNet: Using Landmarks for Better Prediction of 3D Shape
Elena Sizikova

Abstract
Single image 3D shape estimation is arguably one of the most critical problems in modern computer vision, with applications ranging from medical imaging to fashion, design to robotics. However, addressing this problem typically requires learning a complicated nonrigid deformation space, which poses a challenging question of how to preserve the control structure of the underlying shape. In turn, input images are often annotated with complementary information in the form of salient landmarks which can provide additional guidance of viewed shape structure. We propose a novel methodology to incorporate these important landmarks into the shape reconstruction process, and show that the proposed algorithm delivers a higher predictive performance.


Monday, April 23, 2018
Representing Images with Semantic Histograms
Kyle Genova

Abstract
We propose a novel image representation: a normalized histogram containing the relative frequencies of the semantic categories in the image plane. Our representation is low dimensional, sparse, human interpretable, human manipulable, and induces a well-defined metric space based on the earth mover's distance. We establish that this representation offers practical benefits not exhibited by existing alternatives, including representations derived from CNN latent activations. We demonstrate our representation integrates effectively with existing algorithms, by showing it is learnable and that it is an effective conditioning vector for image generation. We conclude by demonstrating that the associated metric space supports meaningful image interpolation and retrieval. Note: This is a presentation on a work in progress.


Monday, April 30, 2018
ExplainGAN: Model Explanation via Decision Boundary Crossing Transformations
Nathan Silberman

Abstract
We introduce a new method for interpreting computer vision models: visually perceptible, decision-boundary crossing transformations. Our goal is to answer a simple question: why did a model classify an image as being of class A instead of class B? Existing approaches to model interpretation, including saliency and explanation-by-nearest neighbor, fail to explicitly illustrate examples of transformations required for a specific input to alter a model's prediction. On the other hand, algorithms for creating decision-boundary crossing transformations (e.g., adversarial examples) produce differences that are visually imperceptible and do not enable insightful explanation. To address this we introduce ExplainGAN, a generative model that produces visually perceptible decision-boundary crossing transformations. These transformations provide high-level conceptual insights which illustrate how a model makes decisions. We validate our model using both traditional quantitative interpretation metrics and introduce a new validation scheme for such an approach.


Thursday, May 17, 2018
Dima Damen