Issue regarding reproducibility

[oonat]

Dear authors,

Firstly, I want to express my gratitude to you for sharing the implementation of your paper. As a master's student working on the multiview object association problem, I found the way you leverage graph neural networks to solve the multiview association problem intriguing and inspiring.

However, I have been having some problems with reproducing the results shared in the paper. I followed the steps provided at README.md, to set up and run the model. However, I couldn't generate the results given in TABLE IV using the pre-trained weights provided at the repo.

These are the results I obtained when I ran python main.py --ConfigPath "config/config_inference.yaml":

P= 0.6745595830281866
R= 0.13732999305407187
F= 0.2170423040655877
TP= 18977
FP= 5278
FN= 98039
TN= 417790
Rand index mean = 0.19834725158713384
Mutual index mean = 0.2373114323871309
homogeneity mean = 0.9472524839589309
completeness mean = 0.610589240404209
v_measure mean = 0.7409091849790294
Mean prec 0 = 99.005546544556
Mean prec 1 = 13.732999305407185

If you can direct me how to reproduce the results in the paper, I would appreciate it.

Thanks.