Assignment -
Need to present face symmetry via a color map on a face data base, and redo the same protocol with normalization using warpping techniques like the attached figure, but just in 2D faces.
Project phases -
1. State of the art related to face asymmetry measurements (metrics), for both 2D and 3D face images.
|
Reference
|
Year
|
Journal /conference
|
Metric
|
2D/3D
|
|
[xxx]
|
|
|
|
|
|
...
|
|
|
|
|
2. Select which of the metrics are appropriate to assess. So far, keep only 2D metrics.
3. Using the selected metrics, analyze each image provided in the Drug addiction database. For this purpose, use the following protocols:
Protocol 1: Without pre-processing the face images, apply the selected metrics on each image ( before and after the addiction).
Represent each face using an appropriate color map, as show in the figure.
You have to quantify statistically each asymmetry in %. You can even compare two or three different metrics as shown in the following example table.
|
Metrics
|
Asymmetry without normalization (Before)
|
Asymmetry without normalization (After)
|
Asymmetry with normalization (Before)
|
Asymmetry with normalization (After)
|
|
Metrics 1
|
% (Average ±std)
|
% (Average ±std)
|
% (Average ±std)
|
% (Average ±std)
|
|
Metrics 2
|
% (Average ±std)
|
% (Average ±std)
|
% (Average ±std)
|
% (Average ±std)
|
|
Metrics 3
|
% (Average ±std)
|
% (Average ±std)
|
% (Average ±std)
|
% (Average ±std)
|
Protocol 2: Repeat the same process as described in protocol 1, except that each image should be normalized using warping techniques.
4. Reference - Include here, all the references you found. Please, basically focus on International Journals and International conferences. Keep all the papers in a folder.
Note - All figures are in attached file.
Attachment:- Assignment Files.rar