Thank you for the implementation of PointNet in TF2.
I have tried to apply PointNet to point clouds with some categorical variables alongside the coordinates of the points. My dataset looks similar to the airplane.npy file that you have provided. I have made some changes to dataset_utils.py so that the categorical variables do not undergo rotation (modified version is attached).
I had to make minor changes to model.py to account for the 6 size input vector and train.py to look in the correct directory.
The training and validation sets are made, and the model is created, but an error occurs when starting training. I get the following error for any batch_size or epoch count.
The list goes on for a while and ends at
I could not find the solution to this anywhere else, as I am not even sure of what the issue is?
Could you take a look at it once?
Hi luis-gonzales,
Thank you for the implementation of PointNet in TF2. I have tried to apply PointNet to point clouds with some categorical variables alongside the coordinates of the points. My dataset looks similar to the airplane.npy file that you have provided. I have made some changes to dataset_utils.py so that the categorical variables do not undergo rotation (modified version is attached). I had to make minor changes to model.py to account for the 6 size input vector and train.py to look in the correct directory.
The training and validation sets are made, and the model is created, but an error occurs when starting training. I get the following error for any batch_size or epoch count.
The list goes on for a while and ends at
I could not find the solution to this anywhere else, as I am not even sure of what the issue is? Could you take a look at it once?
Thank you