FaceCell Library EDK

Embedded and mobile facial identification technology

The FaceCell technology, developed on the VeriLook basis, is designed for embedded biometric systems developers. The technology can be used to create solutions based on PDAs, smart phones and other compact devices with integrated cameras. FaceCell cross-platform algorithm offers decent reliability and identification speed for various mobile or embedded devices.

FaceCell is available for integrators as Embedded Development Kits (EDK) for developing a fast and reliable system on embedded or mobile platform.

Advantages of FaceCell

• Easy integration into PDA or smart phones.
• Simultaneous multiple face processing in live video and still images.
• Fast and reliable face template matching.
• Platform-independent algorithm.
• Reasonable prices, flexible licensing and free customer support.

FaceCell Technology and EDK

• FaceCell Embedded Development Kits. FaceCell 1.2 Library EDK is intended for development projects using hardware based on ARM processors. FaceCell 1.2 source code EDK is intended for developers who are going to integrate facial recognition technology into a custom device.… Read more
• System requirements. Recommended processor is at least 400 MHz ARM family CPU. At least 8 MB of memory required for FaceCell code and data arrays. Windows Mobile 2003 and ARM-Linux are supported… Read more
• Reliability testing results and technical specifications. FaceCell algorithm enrolls a face in less than 2 seconds, matches up to 3,000 faces per second and requires 2.3 kB of memory to store a face template… Read more
• Download. FaceCell brochure, FaceCell algorithm demo application for Windows CE and FaceCell EDK 30-day Trial are available for downloading.  Please contact us today to learn more.

Contents of FingerCell Library and Source EDK

FaceCell Embedded Development Kit is based on the FaceCell embedded facial recognition algorithm that is designed to be used in handheld devices with embedded cameras, such as PDAs or smart phones. FaceCell EDK includes libraries for Linux and Microsoft Windows Mobile on the ARM platform.

It is possible to use FaceCell technology on other platforms and with other operating systems by obtaining the FaceCell source code EDK. The FaceCell algorithm source code is written in ANSI C, thus it is easily portable.

These types of FaceCell 1.2 EDK are available:

• FaceCell 1.2 Library EDK
• FaceCell 1.2 source code EDK

The table below compares different types of FaceCell EDK:

FaceCell 1.2 Library EDK

FaceCell 1.2 Library EDK is intended for development projects using hardware based on ARM processors.

FaceCell 1.2 Library EDK includes these components:

• MS Windows Mobile components
  • FaceCell 1.2 library (for Microsoft Visual Studio 2005 with SP1).
  • FaceCell programming sample in Visual C++ 2005 SP1.
  • FaceCell tutorials in C.
• ARM Linux components
  • FaceCell 1.2 library (for ARM-Linux GCC C compiler).
  • FaceCell tutorials in C.
• Documentation.

FaceCell 1.2 source code EDK

FaceCell 1.2 source code EDK is intended for developers who are going to integrate facial recognition technology into a custom device.

FaceCell 1.2 source code EDK contains these components:

• FaceCell 1.2 source code
  • Project for GCC compiler (ARM-Linux platform)
  • Project for Microsoft Visual Studio 2005 (Pocket PC 2003 and Pocket PC 2005^* platforms)
• FaceCell 1.2 Algorithm and Source Code Description
• Sample applications:
  • Project for GCC compiler (ARM-Linux platform)
  • Project for Microsoft Visual Studio 2005 (Pocket PC 2003 and Pocket PC 2005^* platforms)
• FaceCell EDK developer's guide

^*Pocket PC 2005 development requires Windows Mobile 5.0 SDK for Pocket PC.

FingerCell EDK System Requirements

• ARM-based 400 MHz processor is recommended for face enrollment in less than two seconds. Supported ARM processor core families are: Cortex, XScale, StrongArm, ARM11, ARM10, ARM9.
• At least 8 Mb of memory for FaceCell code and data arrays.
• ARM Linux (glibc 2.3.4 or later) or Microsoft Windows Mobile 2003 (or later) operating system
• (Optional) Embedded camera with at least 320 x 240 pixels physical resolution (640 x 480 pixels recommended). The device must be running Microsoft Windows Mobile 5.0 to use the embedded camera with the demo application.

Please note that FaceCell source code EDK can be easily ported to most other platforms and processors.

Reliability and Performance Test Results

All tests were performed on HP iPAQ with XScale PXA270 processor running at 416 MHz

The FaceCell technology is intended for hardware with lower computational capabilities than PCs. Compared to the PC-based VeriLook 4.0 algorithm, the FaceCell 1.2 algorithm has a higher, but acceptable False Rejection Rate. The graphical chart compares FaceCell 1.2 ROC with VeriLook 4.0 ROC using face images from ROC I subset of the FRGC database.

Click to zoom

Experiment 1 and Experiment 2 were performed with the face images according to FRGC protocol:

• Experiment 1 measures performance on the recognition from frontal facial images taken under controlled illumination. The biometric samples in the target and query sets consist of a single controlled still image in high resolution.
• Experiment 2 is designed to examine the effect of multiple still images on performance. The biometric samples in the target and query sets are composed of the 4 controlled images of each person from a subject.

See Overview of the Face Recognition Grand Challenge (PDF) for more details on FRGC experiments protocol.

Notes:

• Part of images in the FRGC database is 1600 x 1200 pixels, and the other part is 2272 x 1704 pixels. The technical specifications for FaceCell algorithm are given for 640 x 480 pixels images.
• No score normalization techniques were applied while calculating these ROC curves, although FRGC protocol allowed to use score normalization.

All face templates should be loaded into RAM before identification, thus the maximum face templates database size is limited by the amount of available RAM.