search

charlesq34 / pointnet

PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation
Other
4.76k stars 1.45k forks source link

How to use the model file .ckpt for prediction? #224

Open protactinium91 opened 4 years ago

protactinium91 commented 4 years ago

How can I use the .ckpt file after I have trained model? I would like to test it for another dataset, to see if the trained model actually works

syenpark commented 4 years ago

I believe you can make use of this: https://cv-tricks.com/tensorflow-tutorial/save-restore-tensorflow-models-quick-complete-tutorial/. Or you can reference evaluate.py in this github by tracking parser.add_argument('--model_path', default='log/model.ckpt', help='model checkpoint file path [default: log/model.ckpt]').

AliveForU commented 4 years ago

How can I use the .ckpt file after I have trained model? I would like to test it for another dataset, to see if the trained model actually works

Hi,do you solve it?I also got .ckpt file,but don't know how to test another dataset

b-analyst commented 3 years ago

Hi, you have to initialize from that checkpoint. For example, if you're running this with BERT, there is a parameter --init_checkpoint which then you would enter the .ckpt with the highest number.

Lizhinwafu commented 3 years ago

How can I use the .ckpt file after I have trained model? I would like to test it for another dataset, to see if the trained model actually works

Part segmentation:Why can't I use the labeled data to test the output and get the predicted point cloud?

chongma commented 3 years ago

It's shown in test.py