Hi @WayneWWW,
The only difference in the source file is from line 382 to 389:
/* Finally render the osd output */
#ifdef PLATFORM_TEGRA
//transform = gst_element_factory_make ("nvegltransform", "nvegl-transform");
transform = gst_element_factory_make ("queue", "queue");
#endif
//sink = gst_element_factory_make ("nveglglessink", "nvvideo-renderer");
// sink = gst_element_factory_make ("nvoverlaysink", "nvvideo-renderer");
sink = gst_element_factory_make ("fakesink", "nvvideo-renderer");
Regarding the config file you can see that the difference is that I am using the models optimized for Jetson Nano by nvidia.
Both modifications have been done according to what @DaneLLL is suggesting here:
Those modifications are necessary otherwise the Jetson Nano is not able to process the RTSP video stream for it skips a lot of frames.
Thanks again!
ps. I am using only 1 RTSP video source 640*480 resolution at 15FPS.