How to create tf record for deepstream custom training for ALPR model

Accelerated Computing Intelligent Video Analytics DeepStream SDK
1

Please provide complete information as applicable to your setup.

• Hardware Platform (Jetson / GPU): Jestson Nano
• DeepStream Version: 5.1
• JetPack Version: 4.5
• TensorRT Version: 7.1.3.0
• TLT Version: 3.0

Hello there,

We are trying to train the custom ALPR model and we are facing this error:
Cmd:

tlt detectnet_v2 dataset_convert -d ./SPECS_tfrecord.txt -o ./lpd_tfrecord/lpd

Error:

“No file found at: {}. Did you run docker login?”.format(config_path)
AssertionError: Config path must be a valid unix path. No file found at: /home/amnt_ssip/.docker/config.json. Did you run docker login?

Please help us to solve this issue.

Thanks and Regards,
Naresh Mungpara

3

Hi,

Do you run the tlt command on Jetson?
Please noted that training with TLT is only on x86 with NVIDIA GPU such as a V100.

Thanks.

4

Yes, we have run it on Jetson.

Is there any other workaround except training it on V100, can another GPU be used?

5

Hi,

You will at least need a dGPU on the x86 system.

V100 is recommended since we don’t test all the GPUs for training.
But in general, a desktop GPU should work.

Thanks.