ranahanocka
commented
5 years ago Hi, I am not sure I understood the collapse criteria, is it determined completely outside of the network? In other words, is this collapse order determined regardless of the learned features ?
Anyway, yes you are correct, you should modify the code in models/layers/mesh_pool.py and specefically, you should sort the edges based on the criteria you determined in the __build_queue method. Right now we use a heap implementation for speed since it is dynamic, but if your collapse is static (i.e., it is determined outside of the network) the heap is not necessary. Right now , the heap returns a list of edge_ids which were sorted by the minimum edge magnitude.
To do minimum changes to the code and just get it running as you like (not necessarily efficiently), you should build a list of edge_ids in the priority you wish to collapse them.
For a more efficient implementation, you can add a property to the mesh class which has an edge priority list (for example, in mesh_prepare.py you can add an attribute called edge_priority, which the mesh_pool will use to pool edges)
Goodluck & let me know how it goes :)
zideliu
wofsauge
Hi, I want to change how MeshCNN collapses the edges in order to train it to reduce a mesh based on its normals. The smaller the difference between the input Triangle Normal and the Output, the more likely it is that the edge gets collapsed. The overall goal of me is to write a neural network that reduces the polygon count of any input mesh based on a set of criteria.
Where exactly can i find the code that decides whether or not a neuron should collapse an edge or not? i would guess that its in models/layers/mesh_pool.py and there somewhere in the "__build_queue" function. But since the code is not very well documented, i dont quite understand what its doing there. Any idea where i should try to implement that feature?