Please provide complete information as applicable to your setup.
• Hardware Platform (Jetson / GPU) GPU t4
• DeepStream Version 7
• JetPack Version (valid for Jetson only)
• TensorRT Version deepstream instalation giude version
• NVIDIA GPU Driver Version (valid for GPU only) 535
• Issue Type( questions, new requirements, bugs)
• How to reproduce the issue ? (This is for bugs. Including which sample app is using, the configuration files content, the command line used and other details for reproducing)
• Requirement details( This is for new requirement. Including the module name-for which plugin or for which sample application, the function description)
Hi i tried facedetect in deepstream-test-3.py and i get 1200 fps ! (60 videos whitch each of them is 20 fps)but when i change my input to rtsp i get a lot of pixel distorarion with 9 fps in every 10 stream
I tried with different batched-push-timeout and different batch-size but it doesn’t helf i even tried solutions in here
But i still have that problem
I even used nvtop and it only get 50% gpu process and low vram so why i have this problem and how can i fix it?
The distortion is caused by the packet lost during RTSP transferring. Please improve your ethernet quality. You may also try the TCP protocol with your rtsp stream. rtspsrc (gstreamer.freedesktop.org)
But I can do almost 100 streams with my server without using deepstream and none of them have this problem.
You may try to set the tcp protocol and set longer “latency” with rtspsrc. rtspsrc (gstreamer.freedesktop.org)
I used urisrcbin(in python deepstream-test-3) and only set uri parameter
Ty for responding it was so fast😅
Which propperty did you mean? Drop-on-latency (in rtspsrc) Or latency(in urisrcbin)? because urisrcbin already have latency but i don’t even set it
Int your messages i saw that you said set latency longer
how long did you think is ok (i didn’t set it)?
Another question: is source element rtspsrc in image that i upluad
?
Depends on your situation, you may try with your app.
No. The child elements can be got inside the “child-added” callback. GstChildProxy (gstreamer.freedesktop.org). You need to find out the “rtspsrc” element and set the properties. It is just a common GStreamer usage, not special for DeepStream. You can google for more references. GStreamer: get child element from GstElement* - Stack Overflow
Ty for responding
I do it but it doesn’t even help me it got even worse
It’s not in the picture but i try latency too! In values 2000,20000,1000,6000,4000
The element name of rtspsrc is “rtspsrc”.
I used uridecodebin and it works well after a little modification that i upload in the last message but it is not realtime
I think you didn’t check my coee correctly because my code is python and i print the element in the right
I know but there isn’t any element with than name as a child
It fix the problem by some delays
But the problem is this:
I can have 40 local video with 30 fps with used 100% gpu process(1200 fps at all ) in t4 using facedetect model!
But i can only have 6 rtsp input with 20 fps witch used 50% gpu (120 fps at all)😑
I don’t know why it takes 50% gpu process can you help me? It have to be atleast half of the above
(When u add more rtsp my fps reduced massively)
Live streams can’t work ASAP for the video speed will be impacted by the RTSP server and the ethernet transferring.
This has nothing to do with it because i can set latency to 20,000 and still i can only add 6 rtsp input !!!
Are the 6 RTSP streams from different RTSP sources(servers)?
What will happen if you add the 7th RTSP stream?
I tried both of them
One time same rtsp and another time different rtsp
If i add more my fps decrease massively from 25 to 15
Have you set the nvstreammux parameters correctly with your RTSP streams? DeepStream SDK FAQ - Intelligent Video Analytics / DeepStream SDK - NVIDIA Developer Forums
Can you show us the “nvidia-smi dmon” log when you run 7 or more RTSP streams?
Can you give us the complete pipeline and the parameters you are using?
