Please provide complete information as applicable to your setup.
Environment Details:
• Hardware Platform (Jetson / GPU) : dGPU (x86_64)
• DeepStream Version : 7.1
• JetPack Version (valid for Jetson only) : N/A
• TensorRT Version : 10.3.0.26
• NVIDIA GPU Driver Version : 560.35.05
• Issue Type : Question
I’m building a simple pipeline in Graph Composer and would like to log all metadata generated by the inference models (PGIE, SGIE) to a file. I’ve gone through the extension development documentation, but the recommended workflow for this use case still isn’t entirely clear.
At this point, my aim is to dump the raw metadata into a log file, ideally by developing a custom extension or plugin. However, I’m unsure how to proceed or which components are needed.
Could you please advise on the following:
- Is a custom codelet or extension the right approach for capturing and logging inference metadata?
- What’s the best way to access and filter metadata, especially for relevant detections?
- Are there any reference examples that show how to log metadata to a file in a custom format?
Any suggestions would be greatly appreciated.