search

WangYueFt / dgcnn

MIT License
1.62k stars 421 forks source link

Any idea of the get_edge_feature function for unknown batch size #6

Closed leowang16 closed 5 years ago

leowang16 commented 5 years ago

Hi, thanks for your helpful code. But just one painful question: I would like to integrate your code into my Keras implementation. The function of get_edge_feature is using batch_size, which is 'None' for Keras (dynamic for any batch_size). Is it possible to generate the point_cloud_neighbors without using your solution (point_cloud is reshaped to (-1, num_dims)). I guess tf.scan could work. But I am still struggling with that. Many thanks!

leowang16 commented 5 years ago

I think dynamic shape could solve that. batch_size = tf.shape(point_cloud)[0]

Not-IITian commented 5 years ago

Hi,

@Lebo-Wang Were you able to fix this issue? in Particular, can I decide batch size at run time and not a apriori?

Many thanks for your help

leowang16 commented 5 years ago

batch_size = tf.shape(point_cloud)[0]

Hi,

@Lebo-Wang Were you able to fix this issue? in Particular, can I decide batch size at run time and not a apriori?

Many thanks for your help