CorrI2P
CorrI2P copied to clipboard
rsy6318
About success rate of CorrI2P
Hi Siyu, thanks for your nice work. I noticed that the paper seems to evaluate samples with rte and rre less than a certain threshold. Is it convenient for you to provide the quantitative results of the full sample and the success rate of RansacPnP? Looking forward to your reply!
Because of the lack of storage, I deleted the mediate results. You could retrain the network using the dataset I provided. Fig. 8(b) shows the relationship between recall/success rate and RRE/RTE threshold.
Because of the lack of storage, I deleted the mediate results. You could retrain the network using the dataset I provided. Fig. 8(b) shows the relationship between recall/success rate and RRE/RTE threshold.
I tried to train the model on Kitti, but there're something wrong.
How can I fixed it?
'node_a_num' in the dataset is different from that of option. Change it in the dataset from 256 to 128.
'node_a_num' in the dataset is different from that of option. Change it in the dataset from 256 to 128.
Another bug occurred when training.
I think you need to distinguish between “loss value” and “loss function” in train.py.
由于存储空间不足，我删除了中介结果。您可以使用我提供的数据集重新训练网络。图8（b）显示了召回率/成功率与RRE/RTE阈值之间的关系。
我试图在 Kitti 上训练模型，但有些不对劲。我该如何修复它？
I corrected it and reported a new error, why is that?
数据集中的“node_a_num”与选项中的不同。在数据集中将其从 256 更改为 128。
训练时又出现了一个bug， 我觉得你需要在train.py中区分“损失值”和“损失函数”。
请问你是怎么解决的这个问题 我也遇到了
all number change into 256 or 128 then you will find it sloved perfectly.
发自我的iPhone
------------------ Original ------------------ From: LoveMYChen @.> Date: Tue,Jul 23,2024 2:16 PM To: rsy6318/CorrI2P @.> Cc: crzzx1 @.>, Comment @.> Subject: Re: [rsy6318/CorrI2P] About success rate of CorrI2P (Issue #6)
所有数字都变成 256 或 128，然后你会发现它完美解决了。 自制的 iPhone … ------------------ 译文 ------------------ 来自：LoveMYChen @.** > 日期：2024年7月23日星期二下午2:16 致：rsy6318/CorrI2P @.> 抄送：crzzx1 @.>，评论 * @.***> 主题：回复：[rsy6318/CorrI2P] 关于 CorrI2P 成功率 (问题#6 )
Sorry, he still hasn't solved the AttributeError: 'Tensor' object has no attribute 'deschloss'. If you have runnable code, can you send me a copy