I'm curious about how the F1 score was calculated in the paper.

NVlabs / neuralangelo

Official implementation of "Neuralangelo: High-Fidelity Neural Surface Reconstruction" (CVPR 2023)

https://research.nvidia.com/labs/dir/neuralangelo/

Other

4.27k stars 383 forks source link

I'm curious about how the F1 score was calculated in the paper. #193

Open hanjoonwon opened 3 months ago

hanjoonwon commented 3 months ago

I'm curious about what threshold value was used to determine positives. Also, for models that obtain mesh results from RGB images, the size of the output is often inaccurate compared to real scale. I'm wondering how this was adjusted during result evaluation