ViPPET 2025.2#

Major features and improvements#

New graphical user interface (GUI): Interactive visual representation of pipeline graphs with graphical parameter inspection and modification.
Pipeline import and export: Share and version control pipeline configurations across environments.
Backend and frontend separation: Independent development with fully functional REST API for automation and direct access.
Extensible architecture for dynamic pipelines: Support for custom pipeline types without modifying core components.
POSE model support: POSE estimation models integrated into pipeline configuration.
DL Streamer Optimizer integration: Automatic optimization of GStreamer-based pipelines.
Model management enhancements: Add and remove supported models directly through the application.

This section covers additional details on the new ViPPET’s functionality.

A visual representation of the underlying gst-launch pipeline graph is provided, presenting elements, links, and branches in an interactive view.
Pipeline parameters (such as sources, models, and performance-related options) can be inspected and modified graphically, with changes propagated to the underlying configuration.

Pipelines can be imported from and exported to configuration files, enabling sharing of configurations between environments and easier version control.
Exported definitions capture both topology and key parameters, allowing reproducible pipeline setups.

The application is now structured as a separate backend and frontend, allowing independent development and deployment of each part.
A fully functional REST API is exposed by the backend, which can be accessed directly by automation scripts or indirectly through the UI.

The internal architecture has been evolved to support dynamic registration and loading of pipelines.
New pipeline types can be added without modifying core components, enabling easier experimentation with custom topologies.

Integration with the DL Streamer Optimizer has been added to simplify configuration of GStreamer-based pipelines.
Optimized elements and parameters can be applied automatically, improving performance and reducing manual tuning.

Supported models can now be added and removed directly through the application.
The model manager updates available models in a centralized configuration, ensuring that only selected models are downloaded, stored, and exposed in the UI and API.