Triton Server Integration with DeepStream

[Darshcg]

Hi @marcoslucianops,

Thanks for your Projects. It helped me a lot honestly.
Actually, I have run the Yolov3 model(trained on my Custom Dataset) on Jetson Nano using DeepStream with 4 Cameras. Next, I want to Integrate Triton Server with DeepStream for the same model.
So, my doubts are:
1.) How to do the Integration, what all I need to do extra?
2.) Can I serve the TRT models with the triton Server integrated with DeepStream?

Thanks

[marcoslucianops]

Hi, I have no experience with Triton Server, sorry.

[Darshcg]

Hi @marcoslucianops,

Thanks for your reply.

One more doubt is how to run Inference on Two Models using DeepStream.
I want to run both Yolov3 and face Detection model using DeepStream. What are the steps to be followed?

Face Detection model used: https://github.com/biubug6/Face-Detector-1MB-with-landmark

I followed https://github.com/marcoslucianops/DeepStream-Yolo/blob/master/multipleInferences.md, but I have two Different Model(one is Yolov3 and Face Detection model).

Thanks,
Darshan

[Darshcg]

Looking forward to your reply.

[marcoslucianops]

Your Face Detection model is a caffe model or TensorRT converted model?

[Darshcg]

Thanks for your reply, it is TensorRT Converted model

[marcoslucianops]

Can you send me your Face Detection model config_infer.txt file?

[Darshcg]

Hi @marcoslucianops, below is the config_infer.txt file for my FD Model

[property]
gpu-id=0
net-scale-factor=0.0039215697906911373
model-color-format=0
model-engine-file=model.engine
labelfile-path=sgie/labels.txt
batch-size=1
network-mode=2
num-detected-classes=1
interval=0
gie-unique-id=2
process-mode=2
network-type=0
cluster-mode=2
maintain-aspect-ratio=1
parse-bbox-func-name=NvDsInferParseCustomCenterNetFace
custom-lib-path=/opt/nvidia/deepstream/deepstream-5.1/lib/libnvds_infercustomparser_centernet.so

[class-attrs-all]
pre-cluster-threshold=0.3
threshold=0.7

And Below is my Deepstream_app_config.txt

[application]
enable-perf-measurement=1
perf-measurement-interval-sec=1

[tiled-display]
enable=1
rows=2
columns=2
width=1280
height=720
gpu-id=0
nvbuf-memory-type=0

[source0]
enable=1
type=1
num-sources=1
gpu-id=0
camera-width=640
camera-height=480
camera-fps-n=20
camera-fps-d=1
camera-v4l2-dev-node=0
cudadec-memtype=0

[sink0]
enable=1
type=2
sync=0
source-id=0
gpu-id=0
nvbuf-memory-type=0

[osd]
enable=1
gpu-id=0
border-width=1
text-size=15
text-color=1;1;1;1;
text-bg-color=0.3;0.3;0.3;1
font=Serif
show-clock=0
clock-x-offset=800
clock-y-offset=820
clock-text-size=12
clock-color=1;0;0;0
nvbuf-memory-type=0

[streammux]
gpu-id=0
live-source=1
batch-size=1
batched-push-timeout=40000
width=1920
height=1080
enable-padding=0
nvbuf-memory-type=0

[primary-gie]
enable=1
batch-size=1
gpu-id=0
gie-unique-id=1
nvbuf-memory-type=0
config-file=config_infer_primary.txt

[secondary-gie0]
enable=1
batch-size=1
gpu-id=0
gie-unique-id=2
nvbuf-memory-type=0
config-file=centerface.txt

[tests]
file-loop=0

[marcoslucianops]

add to [secondary-gie0] in deepstream_app_config.txt and [property] in centerface.txt

operate-on-gie-id=1
# class ids you want to operate: 1, 1;2, 2;3;4, 3 etc
operate-on-class-ids=0

[Darshcg]

Added it, and worked Perfectly!
Thank you for your help

[Darshcg]

@marcoslucianops,

When I am inferencing Yolov3 using Deepstream with Triton Server on Jetson nano, the detections are going wrong(as shown in te image below)

I don't know what's going wrong. But when I am using Local Deepstream, I am getting the correct detections.

Can you please help me to resolve the issue?

Thanks,
Darshan

[marcoslucianops]

I think it's a problem/bug on Triton