Final Project (draft)

In this Section:

For the final project, we will be working with 3D facial scan data in collaboration with Prof. Heather Edgar and Kate Rusk of UNM Anthropology.

The data consists of 3-D facial scan data for approximately twenty different subjects. All data is available in a standard VRML format that allows (relatively) straighforward parsing or extraction of the key data elements. In addition, our colleagues in Anthropology have kindly provided manually-annotated landmark coordinates for each image, to enable (much) simpler image registration. Finally, three different labels are available for each target image.

New! Updated label data is now available.

Labels data is now available! This is currently partial data (i.e., we don't have observed values for all fields for all subjects). We will expand it as more data becomes available later. See the Data Formats page for details of the contents of this file.

Turnin keys for the project parts are now available:

Assignment Due date Turnin key
Parts 1 & 2 (registration/variance est) Thurs, April 5 cs429-529.fpP12
Proposal Tues, April 10 cs429-529.fpProp
Presentation (e.g., slides) By presentation date cs429-529.fpPres
Final report Mon, May 7 cs429-529.fpReport

The full landmark data for the registration/variance estimation data is now available.

Data and landmarks (complete) for the registration and variance estimation tasks (Goals 1 and 2) are now available.

Data and landmarks (full) for the feature detection tasks (Goals 1 and 2) are now available.

The goals of this work are to:

  1. Register the multiple scans into the same coordinate space and make them all comparable.
  2. Estimate the measurement noise (variance) in the scanning process. (Due to either subject motion or scanner noise.)
  3. Extract generally descriptive 3-D structural features.
  4. Extract class-discriminitive 3-D structural features.

For the first two parts, there is a base procedure that I recommend all students in the class follow. (Though I am open to suggestions for an alternate procedure, if people have good ideas and justifications for them.) For the second two parts, 429 section students will follow a set procedure, while 529 students are expected to innovate.

Timeline

Tues, Mar 20, 2012
Final project description (draft) handed out
Thurs, Apr 5, 2012
Parts (1) and (2) (image registration and variance estimation) due.
Tues, Apr 10, 2012
(529 section) Proposal due
Apr 26-May 3
(529 section) Final project oral presentations (in class); times to be selected by lottery in class
Mon, May 7, 2012
(All students) Final report due