search

charlesq34 / pointnet2

PointNet++: Deep Hierarchical Feature Learning on Point Sets in a Metric Space
Other
3.17k stars 902 forks source link

KeyError: 'FarthestPointSample' - When trying restore the pre-trained model #204

Closed vajrag closed 4 years ago

vajrag commented 4 years ago

Hello,

When I am trying to restore/load the pre-trained model, I get a KeyError: 'FarthestPointSample'. I am not really sure why this is happening.

I am using the code below to restore the model. I have used pointnet2_cls_msg and pointnet2_cls_ssg models which gives the same error. saver = tf.train.import_meta_graph("msg_model.ckpt.meta") sess = tf.Session(config=tf.ConfigProto(allow_soft_placement=True, log_device_placement=True)) saver.restore(sess, tf.train.latest_checkpoint("./"))

System specifications are: tf-gpu = 1.14.0(both for training and restoring) cuda = 11.0 python= 3.6

I appreciate any hints/solutions to resolve this issue. Thank you

irisliucy commented 3 years ago

@vajrag Have you solved the issue? I have the same issue and would appreciate some pointers.

Thank you