Hello AI World, train network

I use CSI Sony IMX477 Camera, Jetpack 4.6,
Python 3.6.9.
I just want to train a network to detect the objects.
I have carried out all the steps that are in Hello AI World.
with the command
$ camera-capture csi://0
I could run my camera to have photos.
I train my network with the command
($ python3 train_ssd.py --dataset-type=voc --data=data/Features --model-dir=models/Features --batch-size=2 --workers=1 --epochs=1).

then I convert my pretrained network to onnx by the command
($ python3 onnx_export.py --model-dir=models/Features)

but when i excute the last Command
($ detectnet --model=models/Features/ssd-mobilenet.onnx --labels=models/labels.txt --input-blob=input_0 --output-cvg=scores --output-bbox=boxes csi://0) i get a problem:

gstDecoder – failed to retrieve next image buffer.
failed to capture video frame.

can anyone help me please

Hi @yazan.doha, it appears that your nvargus daemon needs reset (it provides the interface to the CSI camera) - to do that, you can either reboot your board or run this command:

$ sudo systemctl restart nvargus-daemon

After that, you can try running video-viewer csi://0 to make sure that you can view your camera again.

1 Like

Hi dusty_nv,
thank you very much. Now my camera woks well.

Hi Yazan, ok great, glad you were able to get your camera working again!

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