Why i cant see the labels on the video

• Hardware (GTX1650)

Hello, I want to model with facenet and emotionnet.
This model will first detect the face with facenet and then detect emotion with emotionnet.

However, for this, I learned that I must first use facial landmarks between facenet and emotion net. Based on the deepstream-test2 application, I edited the pgie config file for facenet , facial_landmarks for sgie1 and emotionnet for sgie2.

The model is running but cannot perform classification operations even though it detects the Face.

There is no problem with the log output. and I’m sure the tag files are correct.

What do you think could be the problem? Many thanks in advance for your help.

Please provide complete information as applicable to your setup.
• Hardware Platform (Jetson / GPU)
• DeepStream Version
• JetPack Version (valid for Jetson only)
• TensorRT Version
• NVIDIA GPU Driver Version (valid for GPU only)
• 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)
• The pipeline being used

• Hardware Platform (GTX 1650 GPU)
• DeepStream Version 6.1
• TensorRT Version 8.5.2.2-1+cuda11.8
• NVIDIA GPU Driver Version 510.108.03
• 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)

cfg file for facenet

[property]
gpu-id=0
net-scale-factor=0.0039215697906911373
tlt-model-key=tlt_encode
tlt-encoded-model=…/…/models/tao_pretrained_models/facedetectir/resnet18_facedetectir_pruned.etlt
labelfile-path=labels_facedetectir.txt
int8-calib-file=…/…/models/tao_pretrained_models/facedetectir/facedetectir_int8.txt
model-engine-file=…/…/models/tao_pretrained_models/facedetectir/resnet18_facedetectir_pruned.etlt_b1_gpu0_int8.engine
input-dims=3;240;384;0
uff-input-blob-name=input_1
batch-size=1
process-mode=1
model-color-format=0

0=FP32, 1=INT8, 2=FP16 mode

network-mode=1
num-detected-classes=1
interval=0
gie-unique-id=1
output-blob-names=output_bbox/BiasAdd;output_cov/Sigmoid
cluster-mode=2

#Use the config params below for dbscan clustering mode
#[class-attrs-all]
#detected-min-w=4
#detected-min-h=4
#minBoxes=3
#eps=0.7

#Use the config params below for NMS clustering mode
[class-attrs-all]
topk=20
nms-iou-threshold=0.5
pre-cluster-threshold=0.2

Per class configurations

[class-attrs-0]
topk=20
nms-iou-threshold=0.5
pre-cluster-threshold=0.4

#[class-attrs-1]
#pre-cluster-threshold=0.05
#eps=0.7
#dbscan-min-score=0.5


cfg file for facial_Landmark

[property]
gpu-id=0
int8-calib-file=/opt/nvidia/deepstream/deepstream-6.1/sources/deepstream_python_apps/apps/Model1/int8_calibration.txt
tlt-encoded-model=/opt/nvidia/deepstream/deepstream-6.1/sources/deepstream_python_apps/apps/Model1/FacialLandmarks.etlt
tlt-model-key=nvidia_tlt
model-engine-file=/opt/nvidia/deepstream/deepstream-6.1/sources/deepstream_python_apps/apps/Model1/FacialLandmarks.etlt_b30_gpu0_fp32.engine
#dynamic batch size
batch-size=32

0=FP32, 1=INT8, 2=FP16 mode

network-mode=1
num-detected-classes=1
output-blob-names=softargmax,softargmax:1,conv_keypoints_m80
#0=Detection 1=Classifier 2=Segmentation 100=other
network-type=100

Enable tensor metadata output

output-tensor-meta=1
#1-Primary 2-Secondary
process-mode=2
gie-unique-id=2
operate-on-gie-id=1
net-scale-factor=1.0
offsets=0.0
input-object-min-width=5
input-object-min-height=5
#0=RGB 1=BGR 2=GRAY
model-color-format=2

classifier-async-mode=1
classifier-threshold=0.2

[class-attrs-all]
threshold=0.0
pre-cluster-threshold=0.2
nms-iou-threshold=0.2
dbscan-min-score=0.2


cfg file for EmotionNet

[property]
gpu-id=0

preprocessing parameters: These are the same for all classification models generated by TAO Toolkit.

net-scale-factor=1

model-color-format=1
batch-size=16

Model specific paths. These need to be updated for every classification model.

labelfile-path=/home/jaypear/Desktop/Genel/Görev_Emotion_Analysis/results/output/Labels_emotion.txt
tlt-encoded-model=/home/jaypear/Desktop/Genel/Görev_Emotion_Analysis/results/output/weights/Emotion.etlt
tlt-model-key=nvidia_tao
model-engine-file=/opt/nvidia/deepstream/deepstream-6.1/sources/deepstream_python_apps/apps/Model1/EmotionNet.etlt_b30_gpu0_fp32.engine
infer-dims=3;300;300 # where c = number of channels, h = height of the model input, w = width of model input
uff-input-blob-name=input_1
uff-input-order=0
output-blob-names=predictions/Softmax

0=FP32, 1=INT8, 2=FP16 mode

network-mode=0

process-mode: 2 - inferences on crops from primary detector, 1 - inferences on whole frame

process-mode=2
interval=1
network-type=1 # defines that the model is a classifier.
gie-unique-id=3
operate-on-gie-id=1
classifier-threshold=0.2
num-detected-classes=6

[class-attrs-all]
pre-cluster-threshold=0.2
eps=0.2
group-threshold=1

command line
python3 deepstream_test_2.py /home/jaypear/Downloads/video.h264

Ok. ı solved my problem.
labels file should be a;b;c;d;e…
but my label file was
a
b
c
d
.
.

But now this model just show last label.
For example if my label file is a;b;c;d;e;
it shows always e.
I didnt edit my config file.
Thanks!

Please run with official github deepstream_tao_apps/apps/tao_others at master · NVIDIA-AI-IOT/deepstream_tao_apps · GitHub to check if it works.

Actually… I solved it :D
I changed net-scale-factor and offset and it’s work.
Thanks!

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.