Please provide complete information as applicable to your setup.
•Hardware Platform (Jetson / GPU):Jetson AGX Xavier
•DeepStream Version:5.0
•JetPack Version (valid for Jetson only):Jetpack 4.4
•TensorRT Version：7.1.3-1+cuda10.2
• Issue Type( questions, new requirements, bugs)
questions/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)
My pipeline is as follows:
4 sourcebins: 4 industrial cameras to take images. Images size:img_width=1920,img_height=1200.
nvstreammux:Input parameters:width=1920,height=1200.
nvinfer:Include primary-gie and secondary-gie, use yolov4 plugin,network input =416*416.batchsize=1.
sinkbin1:communication to PLC by socket. When camera get images,infer result must send to PLC as soon as possible.
sinkbin2:save images for transfer learning,functions refer to “deepstream-transfer-learning-app”.Main config parameters:set save images interval in capture_time_rules.csv like this:00:01,23:50,0:00:01
sinkbin3:display detect result on screen.
I want to apply this deepstream pipeline in industrial robot. I need to detect flaw on every product ,so frame drop is not allowed.The pipeline need to process 25-30 images per second.In the beginning,the pipeline works well.But after more than a thousand frames later,some input streams stuck.For example,camera1 and camera2 take photos,and pipeline process this images normally,camera3 and camera4 also take photos,but pipeline does not process images of these two inputs immediately, they stuck for several seconds. But my industrial robot can not alow delay of several seconds for each product.
So I would like to know what could be causing this phenomenon?Is the processing capacity of the system exceeded?Or some problems with my configuration parameters?