I’m a relative beginner working with the deepstream C sample apps to try to integrate my team’s models in a similar way. I’m hoping to modify deepstream-segmentation to take the semantic segmentation mask shown in the video renderer, make it partially transparent, and display it over the sample image. My hope is to use the same process on the output of my own models.
Is this possible with the structure of the deepstream-segmentation? If so, I’d appreciate some pointers on what libraries to look at to make this possible. If not, are there alternate techniques other than the gstreamer method used in the sample app to display this information?
Thanks!
====>Is this possible with the structure of the deepstream-segmentation?
Yes， you can refer the link below: https://github.com/NVIDIA-AI-IOT/deepstream-segmentation-analytics
=====>are there alternate techniques other than the gstreamer method used in the sample app to display this information?
DeepStream is based on gstreamer. If you don’t want to use gstreamer method, you should get the raw data from gstbuffer and implement by yourself.
The example seems to have an output with the input and the mask shown next to each other. I’m specifically looking to overlay the mask on top of the input. Is this possible with nvsosd? It seems to be the function that’s closest to what I’m trying to do.
There is no update from you for a period, assuming this is not an issue anymore.
Hence we are closing this topic. If need further support, please open a new one.
Thanks
Hi, @lparsons ,What do you mean by “the mask shown next to each other” and “the mask on top of the input”?