Effective October 1, 2012, QUALCOMM Incorporated completed a corporate reorganization in which the assets of certain of its businesses and groups, as well as the stock of certain of its direct and indirect subsidiaries, were contributed to Qualcomm Technologies, Inc. (QTI), a wholly-owned subsidiary of QUALCOMM Incorporated. Learn more about these changes

You are here

Shape Knowledge in Segmentation and Tracking

Mar 25, 2014

In this talk I will detail methods for simultaneous 2D/3D segmentation, tracking and reconstruction which incorporate high level shape information.

I base my work on the assumption that the space of possible 2D object shapes can be either generated by projecting down known rigid 3D shapes or learned from 2D shape examples. I minimize the discrimination between statistical foreground and background appearance models with respect to the parameters governing the shape generative process (the 6 degree-of-freedom 3D pose of the 3D shape or the parameters of the learned space). The foreground region is delineated by the zero level set of a signed distance function, and I define an energy over this region and its immediate background surroundings based on pixel-wise posterior membership probabilities. I obtain the differentials of this energy with respect to the parameters governing shape and conduct searches for the correct shape using standard non-linear minimization techniques.

This methodology first leads to a novel rigid 3D object tracker. For a known 3D shape, the optimization here aims to find the 3D pose that leads to the 2D projection that best segments a given image. I also show how the approach could be accelerated to a point where real time processing on a mobile phone becomes possible.

Next, I explore deformable 2D/3D object tracking. I use a non-linear and probabilistic dimensionality reduction, called Gaussian Process Latent Variable Models, to learn spaces of shape. Segmentation becomes a minimization of an image-driven energy function in the learned space. I can represent both 2D and 3D shapes which I compress with Fourier-based transforms, to keep inference tractable. I extend this method by learning joint shape-parameter spaces, which, novel to the literature, enable simultaneous segmentation and generic parameter recovery. These can describe anything from 3D articulated pose to eye gaze.

Victor Adrian Prisacariu, University of Oxford

Date: 25-03-14
Time: 13:30 -15:00
Location: TU Vienna
Zemanek Lecture Room (Room Number: HHEG01)
1040 Vienna, Favoritenstraße 9-11, Stiege III, ground floor, light green area

Event Location

TU Vienna
Zemanek Lecture Room (Room Number: HHEG01) 1040 Vienna, Favoritenstraße 9-11, Stiege III, ground floor, light green area