Face Tracking Software Development Kit
The Face Tracking Software Development Kit (FT-SDK) provides tools to detect faces, track faces, and find facial landmarks in images using a straightforward C-language interface.
Features
Face Detection |
- Finds faces of any size, from very small (<10 pixels inter-eye distance) to very large
- Estimates 3D head pose from frontal to full profile
- Locates facial landmarks (e.g. eyes, nose, etc.)
|
Face Tracking |
- Maintains unique IDs for visually continuous face tracks
- Retains summary statistics as well as detailed trajectory information for each face track
- Captures highest quality face(s) for each face track
- Visual attention reduces computational time for video processing
- Shot boundary detection prevents faces from erroneously being tracked through scene breaks
|
Capabilities |
- Robust to variations in appearance due to age, gender, ethnicity, glasses, facial hair, etc.
- Robust to variable lighting conditions, background scenery and degraded image/video quality
- Supports real-time video processing from stationary or moving cameras
- Supports arbitrary image/video pixel resolution
- Supports single and multi-threaded processing (for multi-core/multi-CPU platforms)
|
Supporting Modules |
- Image IO Library: provides image read and write support for jpeg, gif, bmp, png, ppm, and pgm image formats
- Video IO Library: provides frame capture from mpeg-1, mpeg-2, mpeg-4 (including H.264), and flash video codecs
|
Sample Programs |
- Provides sample program source code for common use cases:
- Face detection on single images or lists of images
- Off-line video processing on input videos or lists of images
- On-line video processing on input videos or lists of images
- GUI applications for ease of use and immediate evaluation (coming soon)
|
System Requirements
- Microsoft Windows (2000, XP, Vista), Linux, or Intel Mac OS X 10.5
- Intel Pentium 4-Class (Intel Core2 Duo recommended)
- 512 MB RAM (1 GB RAM recommended)
- 150 MB free disk space
Documentation