| |
Face image indexing and following rapid face search IIT Face Indexing SDK (IIT FI SDK) |
| |
Overview:
IIT Face Indexing SDK (IIT FI SDK) - software solution for face image indexing and following rapid face search - search of "closest" face images in the indexed image base. This software product can be of use both for different image search services and for specialized system creation in biometrics, security and surveillance, multimedia data and picture archives processing.
Features:
- Input data: Set of grayscale images from 8x8 to 1024x1024 pixels of size
- Output data: list of "closest" candidates sorted by relevance
- Options:
- Each image may be coded by special key
- Search Speed/Robustness ratio may be tuned by user
- Indexing algorithm is stable relative to additive noise
- Indexing algorithm is invariant relative to brightness and contrast changing
- Indexing speed: up to 10.000 images/sec;
- Search speed: up to 10.000.000 images/sec (Intel Pentium-IV-2GHz);
- Image base amount: up to 1.000.000 images;
- Average size of one indexing record: about 100 byte.
Platforms:
| Platforms | Compiler | Status |
| Intel compatible | Visual C++ (Windows 98, 2000, XP, Server 2003) | + |
| Intel compatible | GNU gcc or compatible (Linux) | + |
DSP TMS320C6000 Family DSP TMS320C5500 Family | Code Composer Studio (Texas Instruments) | Contact us |
| ARM7 Architecture | ARM Development Tools | Contact us |
| Other Processors | | Contact us |
Recommended hardware configuration:
- CPU: Pentium III 500MHz
- Memory: 128MB
Advantages:
- High speed of indexing and searching
- Small size of indexing record
- Robustness relative to brightness, contrast and noise
- Convenient API. Easy incorporation of face recognition component into the user's applications
- Ready "hardware+software" solution based on "Intellect" platform (including cameras, frame grabbers and multi-channel video surveillance software "Intellect" with IIT Face Recognition module incorporated) from ITV company
|
|