Question about the eval results #13
Description
Hello! Thank you very much for your test code. However, there are discrepancies between the results I obtained by running the code and those presented in the paper. Could you tell me if there are any other configurations that I need to pay attention to?
On the 7Scenes-Dense benchmark, I achieved test results of 0.064 for average accuracy (avg acc.) and 0.022 for average completeness (avg comp). On the ETH3D benchmark, the corresponding results are 0.52 and 0.31.But in the paper, the results of 7scenes are 0.046/0.029, and ETH3D are 0.414/0.309.