rzhou2197 / pointnet Goto Github PK

Attachment Strategy Learning(Timing and Location)

During the orthodontic process, some attachments would be added to teeth, to add a force point to the tooth surface. The strategy of 
adding attachment contains 3 part of decisions: When to set an attachment, which kind of attachment and which tooth it should be set on. 
That is, orthodontic steps count s(5-15), attachment type c(0-10),  And attachment position p(0-31). triplet <s,c, p> can be view as 
3D coordination in latent space. 

Data: initial 3D model of 32 crowns, an array of triplets<s, c, p> 

Route: Construct a spatial feature learning network that learns from 3D models and the output of deep network is point cloud. 

use python3 train.py to run the model.