chamfer l1 not consistent with paper number (order of magnitude difference)

autonomousvision / occupancy_networks

This repository contains the code for the paper "Occupancy Networks - Learning 3D Reconstruction in Function Space"

https://avg.is.tuebingen.mpg.de/publications/occupancy-networks

MIT License

1.49k stars 291 forks source link

chamfer l1 not consistent with paper number (order of magnitude difference) #61

Closed hua-wu-que closed 4 years ago

hua-wu-que commented 4 years ago

Hi, I ran your pretrained model, but found the chamferL1 is order of magnitude lower than the number reported in the paper. for example, on shapenet chair, the number running your code is "0.020721574735611238", while in the paper it says: "0.228"

m-niemeyer commented 4 years ago

Hi @hua-wu-que , thanks for your interest in the project!

In our project, we we use 1/10 times the maximal edge length of the current object’s bounding box as unit 1. In practice, this means that we multiply the Chamfer-L1 values by a factor of 10. We describe this more detailed in Section 4 - Metrics of the main paper. Good luck with your research!