It's a great work you have done with PointNet ! Thank you for that.
Actually,I have trained my own data( train0[16,2048,3] ,train1[16,2048,3], test0[16,2048,3]) using PointNet.But I found that the "mean loss >1'.And I modified the parameters(bach_size=1,NUM_CLASSES=3,fully_connected=3).
Could you provide some advice about that?
Hello,
It's a great work you have done with PointNet ! Thank you for that. Actually,I have trained my own data( train0[16,2048,3] ,train1[16,2048,3], test0[16,2048,3]) using PointNet.But I found that the "mean loss >1'.And I modified the parameters(bach_size=1,NUM_CLASSES=3,fully_connected=3). Could you provide some advice about that?
Thank you very much in advance.