elchun/contact_graspnet_pytorch

The inferencing results are weird

Closed this issue · 2 comments

Hi, thank for your pytorch implementation of Contact Graspnet. One of the problems I ran into was that the prediction was weird and completely disorganized.
微信图片_20240124111604

Someone had a similar problem with the official TensorFlow version and needed to recompile the "tf_ops" file. What should I do to fix this? Thanks

@CaptainWuDaoKou what did you end up doing to fix this? Thanks!

@CaptainWuDaoKou what did you end up doing to fix this? Thanks!

It's kind of a dumb problem as I forgot to load the model weights.