search

iMoonLab / MeshNet

MeshNet: Mesh Neural Network for 3D Shape Representation (AAAI 2019)
MIT License
345 stars 60 forks source link

Simplify Model to .obj #23

Open atale480 opened 2 years ago

atale480 commented 2 years ago

Can you explain a little bit more about how to convert .off to .obj ? I'm trying to use other dataset ( Princeton Segmentation Benchmark ) in this model. Thanks in advance !!!

yifanfeng97 commented 2 years ago

You can try this function. http://www.open3d.org/docs/0.9.0/tutorial/Basic/file_io.html. Read .off file by open3d and write it in the .obj format. A gentle reminder, you can re-scale and move the center of the 3D object to the origin for better learning on down-stream tasks.

atale480 commented 2 years ago

You can try this function. http://www.open3d.org/docs/0.9.0/tutorial/Basic/file_io.html. Read .off file by open3d and write it in the .obj format. A gentle reminder, you can re-scale and move the center of the 3D object to the origin for better learning on down-stream tasks.

Can you explain how to do re-scale and moving of 3D objects because I have no idea how to work with off file.

atale480 commented 2 years ago

In the Princeton Segmentation Benchmark dataset all mesh files have different number of vertices and faces even for same category. Hence I'm thinking to use this MeshNet to get a representation of all dataset files having same number of features and edge_index so this processed data can be passed to Graph Convolution Network. and one more thing, I'm not finding where the features are stored after running this model on pre-processed ModelNet40 dataset.