RTSP stream with deepstream-gaze-app has big latency

Accelerated Computing Intelligent Video Analytics DeepStream SDK
1

Hello,

We are facing an issue executing the gaze app with rtsp streaming : the app executes correctly but we have a delay about 20 seconds using the command “/deepstream-gaze-app 3 …/…/…/configs/nvinfer/gaze_tao/sample_gazenet_model_config.txt rtsp://192.168.1.168:554/live_st1 ./gaze”.
Although there is almost no delay when we display the rstp video stream using the command “ffplay rtsp://192.168.1.168:554/live_st1” and we have 4-5 seconds delay when we display the usb video with “./deepstream-gaze-app 3 …/…/…/configs/nvinfer/gaze_tao/sample_gazenet_model_config.txt v4l2:///dev/video0 ./gaze”.

Do you know how to solve this issue please ?

Here is our configuration :

Hardware platform : Jetson AGX ORIN DevKit

Deepstream 7.0

Jeckpack 6

TensorRT 8.6.2.3

Cuda 12.2

NVIDIA GPU Driver version 540.3.0

2

Update : If I execute the command “./deepstream-faciallandmark-app 3 …/…/…/configs/nvinfer/facial_tao/sample_faciallandmarks_config.txt rtsp://187.101.100.131:554/live_st1 ./faciallandmark” I have real time inference, no delay. It may be an issue concerning directly the gaze-app and not deepstream

4

compared with faciallandmark sample, there is special gazenet model in gaze sample pipeline. you can remove the second sgie temporarily. if there is no big delay, it should be related to gazenet model processing.

5

Okay thank you

6

Sorry for the late reply, Is this still an DeepStream issue to support? Thanks!

7

Hi ! No, I think it is more like a gazenet related problem