search

xtudbxk / DSRG-tensorflow

a tensorflow version for DSRG (Weakly-Supervised Semantic Segmentation Network with Deep Seeded Region Growing)
75 stars 20 forks source link

Missing checkpoint in trained_model_for_dsrg_tf.tar.gz #8

Closed jsdd25 closed 5 years ago

jsdd25 commented 5 years ago

Hello! Your work is amazing! However, the checkpoint file is not included in the trained_model_for_dsrg_tf.tar.gz. Without it, I can not restore Tensorflow model. Could you please upload it? Thank you very much!

xtudbxk commented 5 years ago

Do you mean the checkpoint files generated by the trained model? You can just train this network and it will generate the checkpoint files by itself.

jsdd25 commented 5 years ago

My computer configuration is not high and the training takes a long time. If it is convenient, could you upload it? Thank you very much!

xtudbxk commented 5 years ago

Oh, I see your question. And you don't need to worry about the checkpoint files. This file just goes to tell the tensorflow which one is the latest model. So just passing the "final-0" as the name of the model is ok for tensorflow.

xtudbxk commented 5 years ago

You can find the usage in the restroe_from_model function of https://github.com/xtudbxk/semantic-segmentation-metrics/blob/master/deeplab_aspp_test.py.

When the checkpoint is not available, just pass the "final-0" as the model_path.

jsdd25 commented 5 years ago

@xtudbxk It's very kind of you. Thank you very much!