RPI ISL Facial Expression Databases

 

Source: this database is collected by our own research group.

Purpose: this database is primarily used for facial action unit recognition.

It consists of two subsets:

o      Frontal-view facial expression database

·       Purpose: this database is primarily used for frontal-view facial action unit recognition. It may also be used for face recognition.

·       Properties:

Properties

Descriptions

# of subjects

10

# of images/videos

42

Static/Videos

Videos

Single/Multiple faces

Single

Gray/Color

Color

Resolution

320 * 240

Face pose

Nearly frontal-view

Facial expression

Various

Description of facial expression

Neutral-apex-neutral

Posed facial expressions

Illumination

Indoor uncontrolled illumination

Accessories

N/A

3D data

N/A

Frame rate

30

Ground truth

AU labels of 14 AUs for each frame

Identifications of subjects

 

o      Multi-view facial expression database

·       Purpose: this database is primarily used for facial action unit recognition under natural head movements. It may also be used for face recognition.

·       Properties:

Properties

Descriptions

# of subjects

8

# of images/videos

40

Static/Videos

Videos

Single/Multiple faces

Single

Gray/Color

Color

Resolution

320 * 240

Face pose

5 face poses: frontal, half-left, quarter-left, quarter-right, and half-right

Facial expression

Various

Description of facial expression

Neutral-apex-neutral

Posed facial expressions under natural head movements

Illumination

Indoor uncontrolled illumination

Accessories

N/A

3D data

N/A

Frame rate

30

Ground truth

Coordinates of pupils for selected images (even frames);

Coordinates of 34 facial feature points (e.g. eyebrow corners and eye corners) for selected frames (usually the images with significant facial appearance/pose changes) for each sequence;

Discrete face pose label for each frame;

AU labels of 15 AUs for each frame;

Identifications of subjects.