DeepFloorPlan_Pytorch

This is a pytorch implementation for the floor plan segmentation on the r2v dataset.

As well as a simple 3D mesh modeling script with the ModelNet dataset.

RESULTS

Segmentation

Method	OverallAcc	Class Average
PAPER CLAIMED	0.90	0.89
HRNet	0.947	0.948
HRNet + OCR	0.959	0.947

Mesh Reconstruction

if you need to train the model, you need to replace path in the bash script / config file to the your directory
ModelNet Dataset and r2v/jp should be placed in the ./data.

TRAIN

bash sourcecode/exp/02/train.sh

INFERENCE (TEST)

bash sourcecode/exp/02/inference.sh

MESH RECONSTRUCTION

bash sourcecode/exp/02/mesh.sh