NvDsInferSegmentationMeta output pointers to GPU memory

[ketronium]

The NvDsInferSegmentationMeta struct contains pointers to segmentation inference outputs on the CPU, are there similar pointers somewhere to access the same or similar attributes on the GPU? I want to pass segmentation outputs to a CUDA kernel with the C++ api and ideally would like avoid having to copy those CPU arrays back to the GPU prior to passing to kernel. Thanks

Running from docker container nvcr.io/nvidia/deepstream:6.1.1-devel
**• Hardware Platform: GPU **
• DeepStream Version = 6.1.1
• TensorRT Version v8401
**• NVIDIA GPU Driver Version 525.105.17 **
• Issue Type questions
• Requirement details nvinfer BatchMeta outputs

[yuweiw]

We currently do not implement this. But you may be try to implement by yourself.
You can refer to our source code: fillSegmentationOutput sources\libs\nvdsinfer\nvdsinfer_context_impl_output_parsing.cpp.