hi.
i need to draw some text on image after decoding. i see simplest cuda examples in mm api. is there examples for drawing text labels with cuda?
Hi @alexcon314, the MM API has the NvOSD interface which can overlay text. You can find it’s header at /usr/src/jetson_multimedia_api/include/nvosd.h
The 02_video_dec_cuda
sample also shows using NvOSD to overlay text on the video stream.
Alternatively, the jetson-inference library has support for CUDA fonts in C++/Python, you can see it along with some examples here:
Hi, dusty_nv. thank you for reply.
Is NvOSD are using GPU engine? I found following comment in docs:
enum NvOSD_Mode
MODE_GPU Selects GPU for OSD processing.
Yet to be implemented
I believe it’s using the VIC (video composting engine), which can be offloaded from the GPU.
The jetson-inference implementation of cudaFont that I linked to uses the GPU/CUDA for rendering text overlays.
I will trying it. thank you.