CYENS Center of Excellence, the research and innovation center for interactive media, intelligent systems and new technologies, announced the launch of the beta version of its innovative image/video pipeline Interface tools DgiStreamer announced. DgiStreamer makes complex image processing effortless by automatically creating image processing pipelines and deploying them on any system (i.e. embedded, GPU, CPU) while supporting all the features offered by other popular Software Development Kits (SDKs) like Nvidia’s GStreamer and Deepstream.
Image processing and computer vision pipelines are still difficult to use, especially when working with dedicated SDKs/APIs. As the complexity of the pipeline increases, so does the amount of manual coding required for implementation and testing. In addition, the developer often has to deal with compatibility problems with the SDK versions and the cross-node pipeline functions. Building a pipeline with the tools currently available is not optimal because a lot of time is spent writing additional code and debugging.
The DeepCamera team of the CYENS Center of Excellence led by Dr. Alessandro Artusi has developed an innovative GUI interface tool that serves as a holistic, standalone pipeline builder compatible with GStreamer and Deepstream. The tool provides an intuitive way to simplify pipeline construction by simply dragging and dropping to connect the different blocks of a pipeline. The tool allows scaling to complex types of configurations and is faster and easier to use than the solutions currently on the market.
When developing DgiStreamer, the focus was on practicality. The beta version of DgiStreamer offers features that further simplify construction. A novel debugging system makes the tool system independent, allowing the application to run on a variety of operating systems while providing a playback video stream of the pipeline output. This is an effective way to monitor, manage and/or debug deployed pipelines, both remotely and locally. The tool is also fully compatible with the previous SDK components and versions.
