What is the proper way to do classification pretrained?

Hi,

I am using dusty-nv inference.
I can do the object detection retrain by calling
-pretrained-ssd= xxx.pth

However, when I try to retrain the classification,
I do not see any pth file, just two compressed files.

checkpoint.pth.tar
model_best.pth.tar

I did unzip them but still cannot find the pth.
May I know the proper way to retain the model?
Thx

Hi,

Do you apply the classification retraining with the similar procedure as below:

Thanks.

Hi,

Yes, I create my own dataset for classification. Training from the beginning is ok.
I just don’t know how to continue the training based on my trained model yesterday for example.
Thx

Hi,

You can find some information below for training on a custom database:

Thanks.

ok ok, I can use resume option, thx

If you need to, go back and collect more data and re-train your model again. You can restart the training from where you left off using the `--resume` and `--epoch-start` flags (run `python3 train.py --help` for more info). Then remember to re-export the model.

$ cd jetson-inference/python/training/classification
$ python3 train.py --model-dir=models/ data/ --resume

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