Dear all,
I encountered an issue when running any test apps in the package:
Command:
sudo deepstream-app -c sources/objectDetector_Yolo/config_infer_primary_yoloV3.txt -i samples/streams/sample_1080p_h265.mp4
Same error also appears when I ran other apps like:
sudo deepstream-app -c /opt/nvidia/deepstream/deepstream/samples/configs/deepstream-app/config_infer_secondary_carcolor.txt
Additional information: I ran it on my remote server and login to it via nomachine (NX protocol). The server is not headless, so when I ran the sample app using the plugged monitor, the app could run successfully.
Command:sudo deepstream-app -c source4_1080p_dec_infer-resnet_tracker_sgie_tiled_display_int8.txt
However, it fails remotely. Is there any way to make it display?
I appreciate any kind help. I have been working on it the entire day…
Thanks
Environment:
Ubuntu18.04, tensorRT7.0.0.11, cuda-10.2, cudnn-7.6.5, GPU: Titan V and deepstream 5.0
→ you are not supposed to run with nvinfer config, you should run with samples/configs/deepstream-app/source*** config.
The server is not headless, so when I ran the sample app using the plugged monitor, the app could run successfully.
Command:sudo deepstream-app -c source4_1080p_dec_infer-resnet_tracker_sgie_tiled_display_int8.txt
However, it fails remotely. Is there any way to make it display?
I appreciate any kind help. I have been working on it the entire day…
→ if you want to make it display, another method is you can use sink type 4 UDPSink for rtsp streaming, then you can see it from remote server using vlc.