After the development month, the release of the Python frame Savant 0.2.4 has been announced. This release aims to simplify the use of nvidia deepstream for machine learning problem-solving. By handling the complex tasks with Gstreamer or FFMPEG, Savant allows developers to focus on building optimized output conveyors using declarative syntax (YAML) and Python functions. The framework supports both accelerators in the data center and EDGE devices such as NVIDIA TURING, AMPERE, HOPPER, Jetson NX, AGX XAVIER, Orin NX, AGX Orin, and New Nano. Savant enables the processing of multiple video streams simultaneously and facilitates the creation of video analytics conveyors ready for working applications using NVIDIA Tensort. The project code is distributed under the Apache 2.0 license.
Three new examples of use have been provided:
- Prediction of age and gender, which demonstrates the use of yolov5-Face and the ability to work with the user attributive attribute model. The example showcases age and gender prediction as well as affine transformations using Opencv-Cuda and Python on the GPU.
- Conditional coding highlights the conveyor that draws on frames and codes only based on a specific request or condition. This example shows how to avoid the irrational use of computing resources by performing coding tasks only when objects are found using a model.
- Processing of multiple RTSP streams showcases a simple conveyor that processes two RTSP stacks simultaneously. This example demonstrates Savant’s ability to handle multiple flows effectively.
New opportunities introduced in this