Is that possible to change the optical flow algorithm when running "deepstream_nvof_test.c"

Accelerated Computing Intelligent Video Analytics DeepStream SDK
1

Please provide complete information as applicable to your setup.

**• Hardware Platform (GPU) RTX 3060
**• DeepStream Version 6.1.1
• TensorRT Version
**• NVIDIA GPU Driver Version (valid for GPU only) CUDA11.7

When I run deepstream_nvof_test.c, it output dense optical flow, any suggestions if I want to output sparse optical flow. As there is no nvof plugin, so can’t direct change optical flow algorithms from source code, so where should I change the optical flow algorithm to get sparse optical flow?

3

do you mean the “grid-size”? currently only the 4x4 grid size is supported. please refer to Gst-nvof — DeepStream 6.1.1 Release documentation

4

Thanks for the reply. I have to questions to double:
1.Is that possible we change the nvof internal optical flow algorithm. As I want optical flow motion vectors (sparse optical flow).
2.As in deepstream 6.1.1, there is only test code in deepstream-nvof-test, how to extract the optical flow ,as there is no source code in plugin for gst-novf.

5
  1. currently the low level lib is not open.
  2. you can use Gst-nvof’s dump-of-meta=1 to dumps optical flow output into a .bin file, please refer to Gst-nvof — DeepStream 6.1.1 Release documentation.
6

There is no update from you for a period, assuming this is not an issue any more.
Hence we are closing this topic. If need further support, please open a new one.
Thanks

8

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.