Fraunhofer Institute for Digital Media Technology IDMTTemporal Video Segmentation tools provide information about the video structure by detecting shot boundaries, representative key frames and visual similarity between multiple shots. The information can be used for video structure analysis, content management, and annotation issues.
Temporal Video Segmentation
Features
- Universally applicable signal analysis framework for video content
- Detection independent from distribution channels and formats
- Applicable for file-based scenarios and stream-based scenarios including real-time requirements
- Result polling methods and detailed error reports
- Segmentation results can be provided as information for further video analysis components, e.g. Video Motion Analysis
Analysis Component
Input: Decoded (uncompressed) video frames, color spaces: YUV 4:2:0 PLANAR, RGB, BGR, BGRA, GRAY, other formats upon request
Results: Measurement values and result reporting
Modules
- Shot Detection: Detects shots and its encapsulating shot boundaries
- Key frame detection: Detects representative key frames belonging to new visual relating information
- Scene detection: Detects scenes while analyzing multiple shots relating to each other
- Coming soon: Transition types, such as wipes, fades, and dissolves
The Temporal Video Segmentation component is being continuously enhanced. If you have specific requirements or would like more information please contact us.
Technical Information
- Analysis methods encapsulated in dedicated modules
- Modular architecture supporting customization to individual requirements
- Components available as C++ libraries for Windows, Linux, and OS X for integration into existing systems, other platforms upon request
- Evaluation versions available upon request
Licensing Information
- Basic yearly license for internal use of licensee
- Reseller license per instance (server) or revenue share (Software as a Service)