3D-Room-Layout-Estimation_LGT-Net / Post-Porcessing.md
zhigangjiang's picture
no message
88b0dcb
|
raw
history blame
917 Bytes

Post-Processing

Step

  1. Simplify polygon by DP algorithm

img.png

  1. Detect occlusion, calculating box fill with 1

img.png

  1. Fill in reasonable sampling section

img.png

  1. Output processed polygon

img.png

performance

It works, and a performance comparison on the MatterportLayout dataset:

Method 2D IoU(%) 3D IoU(%) RMSE $\mathbf{\delta_{1}}$
without post-proc 83.52 81.11 0.204 0.951
original post-proc 83.12 80.71 0.230 0.936
optimized post-proc 83.48 81.08 0.214 0.940

original:

img.png

optimized:

img.png