search

WangYueFt / dgcnn

MIT License
1.62k stars 420 forks source link

Where is the categorical vector? #45

Closed Steve-Tod closed 4 years ago

Steve-Tod commented 4 years ago

Hi, thank you for sharing the code. In your paper, there is a categorical vector for semantic segmentation network.

image

However, I didn't find many descriptions of this vector in the paper or any implementation of this in the code. https://github.com/WangYueFt/dgcnn/blob/69b80a370df4125adb6ac376169de2de76a61d67/tensorflow/sem_seg/model.py#L87 Here, after the max pooling, the global feature is directly repeated and concatenated with point features.

What is this categorical vector? Is it necessary?

Steve-Tod commented 4 years ago

Sorry, I found this vector is for part segmentation.

Another question, in PyTorch version conv2d without bias is used but in TensorFlow version conv2d with bias is used. Why there is such a difference?

WangYueFt commented 4 years ago

Sorry, I found this vector is for part segmentation.

Another question, in PyTorch version conv2d without bias is used but in TensorFlow version conv2d with bias is used. Why there is such a difference?

With batchnormalization, w/ or w/o bias doesn't make any difference.