Retrain the Pretrained Model in Deepstream


Hello all,
I want to retrain a Pre-trained object detection model used in deepstream-image-meta-test code with my own custom dataset (images).
My objective is to increase the accuracy of my model by training it on the custom set of images.
As per my observation, the deepstream-image-meta-test.c code uses the model built on caffe framework.
I wish to retrain the same model used in that code to train with my custom dataset.
Is there a way to do so?
Please guide me the required steps to perform the model training on my custom dataset.


TensorRT Version: 7.1.3-1+cuda10.2
GPU Type: Nvidia Tegra X1
Nvidia Driver Version:
CUDA Version: 10.2
CUDNN Version:
Operating System + Version: ubuntu 18.04 LTS 64 bit
Python Version (if applicable): 3.6.9
TensorFlow Version (if applicable):
PyTorch Version (if applicable):
Baremetal or Container (if container which image + tag):

Relevant Files

Please attach or include links to any models, data, files, or scripts necessary to reproduce your issue. (Github repo, Google Drive, Dropbox, etc.)

Steps To Reproduce

Please include:

  • Exact steps/commands to build your repro
  • Exact steps/commands to run your repro
  • Full traceback of errors encountered


This looks like Deepstream related. We are moving this post to Deepstream forum to get better help.

Thank you.

@spolisetty ok

@Morganh , Seems TAO related question?

@kesong Please help me with this…i require this urgently…


Tried this one but training process stops in b/w
The logs of which is:
output_finetune.txt (10.8 KB)

May I know are you using TAO to train?
Your shared link is not about TAO.

@Morganh I dont have any idea about TAO…can you please explain


For TAO, you can find info in Announcement: TAO Toolkit 3.0-21-11 released.

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