RPMNet
RPMNet copied to clipboard
yewzijian
eval_results
Hello, I would like to ask you
How did I obtain the results under the clean condition in the paper? I used eval. py to verify the results of clean-trained. pth, but it is much worse than the results in the paper. May I ask where did I go wrong? Here are the results I have obtained
Hi, you’ll need to set it to the pre trained checkpoint. Follow the instructions in the readme.
On Fri, 19 Jan 2024 at 1:32 AM, Soed-a @.***> wrote:
hi, i have really problem with resume that is None. how did you run eval?
— Reply to this email directly, view it on GitHub https://github.com/yewzijian/RPMNet/issues/40#issuecomment-1898921707, or unsubscribe https://github.com/notifications/unsubscribe-auth/ADIBP664IHGDOIK7AYRRU33YPFMERAVCNFSM6AAAAABB7RNMGWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQOJYHEZDCNZQG4 . You are receiving this because you are subscribed to this thread.Message ID: @.***>
@wuchengzenb this is indeed weird. Do the other settings (noise/partial) work ok?
这确实很奇怪。其他设置（噪音/部分）是否正常工作？
noise
partial
These two did not meet the expected results. Then I also have a question, which is when I use the same model for validation, if I reduce the num_points, the performance gets better. Why is that, for example, when I change it from the default 1024 to 717? What other arguments within the arguments could affect the results?
1024
717
I don’t know why your performance is worse, to be honest. You should be able to get the reported results without any change in the arguments.
The network is trained with 1024 points (with the exception of partial point clouds setting which is trained with 717). So if anything, I would expect 1024 points to work better.
I don’t know why your performance is worse, to be honest. You should be able to get the reported results without any change in the arguments.
The network is trained with 1024 points (with the exception of partial point clouds setting which is trained with 717). So if anything, I would expect 1024 points to work better.
Thank you, I was able to verify it on my classmate's computer, which is quite strange. Then I made some improvements, here are the results of training for 800 epochs under clean conditions. The CD increased, but the other four metrics decreased. Do you think this is an improvement?Which of these five metrics have higher weights?