Software Ecosystem of Event Cameras
Software Ecosystem
Event-based vision systems rely on a layered software stack that enables developers to interact with hardware, process event streams, and build complete applications.
The Prophesee ecosystem is centered around two complementary offerings:
- Metavision SDK Pro: A complete SDK providing precompiled binaries and full source code for both open and advanced modules.
- OpenEB: An open-source framework (Apache 2.0) exposing the core components of the SDK, enabling customization and support for additional platforms.
Together, these form a flexible environment that supports both research and production deployments.
Overview of the Software Stack
At the core of the ecosystem is OpenEB, which provides access to event data and fundamental processing components.
On top of this foundation, the Metavision SDK adds higher-level modules, applications, and tools for building complete solutions.
The stack can be understood as a set of layers:
- Hardware interaction (Drivers)
- Data access and streaming (HAL, Stream)
- Processing and algorithms (Core, CV, ML)
- Application development (APIs and tools)
This layered architecture allows developers to work at different abstraction levels depending on their needs.
Core Platform: OpenEB and Metavision SDK
OpenEB
OpenEB provides:
- Access to core SDK modules (HAL, Core, Stream, UI, etc.)
- Full source code under the Apache 2.0 license
- Flexibility to adapt the stack to unsupported platforms
- A foundation for research and custom development
Metavision SDK Pro
Metavision SDK Pro extends OpenEB with:
- Advanced modules (CV, ML, Analytics, Calibration)
- Precompiled binaries
- Optimized algorithms for production use
- Additional tools and applications
Drivers, APIs, and SDKs
Drivers
Raspberry Pi Driver
Raspberry Pi platforms are supported through OpenEB and the Metavision SDK without requiring custom driver development.
- Repository: rpi-sensor-drivers
Key Features
- Direct access to event stream data
- Integration via the HAL (Hardware Abstraction Layer)
- Ready-to-use tools for visualization and recording
- Simplified integration for rapid prototyping
NVIDIA Jetson Driver
A custom driver for NVIDIA Jetson platforms is currently under early-stage development.
- Repository: [add link]
Key Features
- Early-stage support for Jetson platforms
- Basic device communication and event streaming
- Integration with OpenEB / Metavision HAL (work in progress)
- Focus on enabling experimentation and future optimization
SDK Modules
Open Modules
(Core functionality available in OpenEB)
- HAL
- Base
- Core
- Core ML
- Stream
- UI
Advanced Modules
(Available in Metavision SDK Pro)
- CV
- CV3D
- Analytics
- Calibration
- ML
APIs
Metavision SDK offers two APIs:
The C++ API provides high performance and fine-grained control for real-time applications.
The Python API enables rapid prototyping and is well suited for experimentation and machine learning workflows.
Visualization and Debugging Tools
Visualization Tools
- Metavision Studio: Advanced GUI for visualization and recording
- Metavision Viewer: Lightweight alternative for streaming and recording
- Metavision XYT: 3D visualization of event data
Metavision Studio
Provides an interactive interface to visualize and record event streams from cameras or files (RAW, DAT, HDF5).
Metavision Viewer
Allows streaming and visualization of events with additional features:
- Region of Interest (ROI)
- Event Rate Controller (ERC)
- Camera configuration management
Debugging Tools
- Metavision Data Rate
- Metavision Platform Info
- Metavision Software Info
- Metavision File Info
- Active Pixel Detection
Calibration Tools
- Camera Focusing
- Camera Calibration Pipeline
- Ground Plane Calibration
Recording and Playback Tools
Recording functionality is available through:
- Metavision Studio
- Metavision Viewer
These tools support both live recording and playback of event streams.
Conversion Tools
- File to CSV Converter
- File to HDF5 Converter
- File to Video Converter
- File to DAT Converter