DeepStream Jetson nano with IP camera

Accelerated Computing Intelligent Video Analytics DeepStream SDK
#1

How to connect to multi ip camera using rtsp in the example objectDetector_Yolo

  • WARN: <create_pipeline:930>: Num of Tiles less than number of sources, readjusting to 2 rows, 1 columns
    Opening in BLOCKING MODE
    NvMMLiteOpen : Block : BlockType = 261
    NVMEDIA: Reading vendor.tegra.display-size : status: 6
    NvMMLiteBlockCreate : Block : BlockType = 261
    ** INFO: <bus_callback:175>: Pipeline running

Creating LL OSD context new
ERROR from source: Could not open resource for reading and writing.
Debug info: gstrtspsrc.c(7469): gst_rtspsrc_retrieve_sdp (): /GstPipeline:pipeline/GstBin:multi_src_bin/GstBin:src_sub_bin1/GstURIDecodeBin:src_elem/GstRTSPSrc:source:
Failed to connect. (Generic error)

[source0]
enable=1
#Type - 1=CameraV4L2 2=URI 3=MultiURI
type=3
uri=rtsp://admin:vnnet123456@192.168.1.202
#uri=file://…/…/samples/streams/sample_1080p_h264.mp4
num-sources=1
gpu-id=0

(0): memtype_device - Memory type Device

(1): memtype_pinned - Memory type Host Pinned

(2): memtype_unified - Memory type Unified

cudadec-memtype=0

[source1]
enable=1
#Type - 1=CameraV4L2 2=URI 3=MultiURI
type=3
uri=rtsp://admin:vnnet123456@192.168.1.204
#uri=file://…/…/samples/streams/sample_1080p_h264.mp4
num-sources=1
gpu-id=0

(0): memtype_device - Memory type Device

(1): memtype_pinned - Memory type Host Pinned

(2): memtype_unified - Memory type Unified

cudadec-memtype=0

#2

Hi,

$ gst-launch-1.0 uridecodebin uri=rtsp://admin:vnnet123456@192.168.1.202 ! nvoverlaysink

Please use DS4.0.1 and check if the uri is valid or not.

#3

Hi. Thank you.
I have connected many ip cameras.
Is there any way to increase performance? ( examole Yolo v3)
Present:
1 camera detection: FPS: 24
2 camera detection: FPS: 12
4 camera detection: FPS: 6

#4

Hi,
Do you run deepstream_app_config_yoloV3.txt or deepstream_app_config_yoloV3_tiny.txt?

#5

I run deepstream_app_config_yoloV3_tiny.txt

#6

Hi,

Please check this comment:
https://devtalk.nvidia.com/default/topic/1064871/deepstream-sdk/deepstream-gst-nvstreammux-change-width-and-height-doesn-t-affect-fps/post/5392823/#5392823

This sample can reach 20 FPS with YOLOv3 on Nano.

Thanks.

#7

My project also gets 20 fps on yolov3-tiny.
but when using multiple cameras, it is reduced.
Is there a workaround?

#8

Hi,

YoloV3 is much complicated than the tiny version.
Try to apply the similar update to the YOLO_tiny and you should get a better result than 20FPS.

Thanks.

#9

I also used yolov2_tiny, the results were improved with 25 fps.
but when using multiple cameras, the result is reduced.

#10

result: 1 cam
https://drive.google.com/file/d/1v_zRRdScmpHF9aIThv0AmwNkdAJUB1z-/view?usp=sharing

result: 4 cam
https://drive.google.com/file/d/1U9pV-O9BCPvwi551-VOxa3PQPkEpCJCc/view?usp=sharing

#11

Hi,

Could you share your config for multi camera with us?

Have you updated the batch size of streammux and primary-gie into 4?
For 4 stream input, the streammux and nvinfer batch size need to be 4 as well.

Thanks.

#12

Hi,
Could you share your config file for multiple cameras ?
I am trying to run yolo3 on multiple cameras, but still no idea how to modify config file

thanks!

#13

Hi,
You have shared the result of running 4 cam. Would be better if you can share the config file of running 4 cam.

#14