What‘s the difference of the final metrics log between test.py and eval.py???

[Wyyspace1203]

As described in the title, I found that the metrics obtained after running test.py are different from those obtained after running eval.py, but I believe they should be the same. Why is this happening?

[huantuoyushui]

hello,Have you figured out why the output of test.py and eval.py is different

[Wyyspace1203]

hello,Have you figured out why the output of test.py and eval.py is different

I don't know yet

[ccjccjccj]

I suppose the results in test.py are the mean values of all registration pairs, and the results in eval.py are the mean values of all scenes (first calculate the mean values of each scene, then calculate the mean values of all scenes)