search

charlesq34 / pointnet

PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation
Other
4.71k stars 1.44k forks source link

Output Labels are not in one-hot encoding for classification #131

Closed sumitsinha closed 6 years ago

sumitsinha commented 6 years ago

The output labels for each point are single numbers between 1-40. The last output layer has 40 output units. I wanted to know should the labels be converted to one hot encoding (vector of 40 values) before giving it to the for model training?

sumitsinha commented 6 years ago

I see we are using sparse_softmax_cross_entropy_with_logits in the loss function that takes care of one-hot encoding problem.