I have a lens and I want to add a custom filter based on gst-launch 1.0 and nvarguscamerasrc. This filter implements a set of image correction algorithms of our company. What should I do to add the filter?
hello 1508723374,
may I have more details,
for example,
(1) what kind of this filter you’re mentioned.
(2) is this a physical device (different lens) or, different software parameter settings.
The final effect I want to achieve is to add a custom filter, such as testfilter, to the following command.
gst-launch-1.0 nvarguscamerasrc sensor-id=0 wbmode=0 saturation=0 blocksize=24834048 ! ‘video/x-raw(memory:NVMM), width=(int)4128, height=(int)3008,format=(string)NV12, framerate=(fraction)29/1’ ! testfilter ! nvv4l2h265enc ! h265parse ! filesink location=cam-0.mp4
hello 1508723374,
it looks like post-processing.
you may refer to Sample Applications, such as 11_video_osd (video osd) for post-processing.
What we need is real-time image processing, which doesn’t quite match our requirements.
I have found an architecture article that satisfies us and we want to modify the nvsample_cudaprocess_src code to implement our own image correction algorithm. Do you think this approach is feasible?
VPI in a GStreamer pipeline - Jetson & Embedded Systems / Jetson Nano - NVIDIA Developer Forums
yes, it may doable, please refer to DeepStream SDK, GStreamer Plugin Overview as see-also.
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.