KeyError: '2021_09_09_13_20_58'

[Chinese-Coding]

KeyError: '2021_09_09_13_20_58'

Chinese

在我训练模型时, 遇到了这样一个错误 (Figure 1), 经过我的调试我发现错误原因为: 在数据集 OPV2V 和 OPV2V-H train文件夹下有时间戳 2021_09_09_13_20_58的数据, 而在 opv2v_4modality.json 和 opv2v_4modality_in_order.json 中均没有该数据. 一个简单的解决办法就是将它们从对应的目录下移除即可, 或者还有一个方法就是在 opv2v_4modality.json 和 opv2v_4modality_in_order.json 加上该时间戳的数据, 但是我认为这一步可能会引起训练精度的问题, 所有没有这么做.

English

When I was training the model, I encountered such an error (Figure 1). After debugging, I found that the cause of the error was: there was data with the timestamp 2021_09_09_13_20_58 in the train folder of the datasets OPV2V and OPV2V-H, but there was no such data in opv2v_4modality.json and opv2v_4modality_in_order.json. A simple solution is to remove them from the corresponding directories, or another way is to add the data with the timestamp to opv2v_4modality.json and opv2v_4modality_in_order.json, but I think this step may cause problems with training accuracy, so I didn't do it.

Figure

[Chinese-Coding]

Additional findings

Chinese

我在 v2xset_4modality.json 中发现了 2021_09_09_13_20_58 的数据, 但是应该是不能使用的, 因为 id 对不上.

English

I found the data of 2021_09_09_13_20_58 in v2xset_4modality.json, but it should not be used because the id does not match.

Figure

[yifanlu0227]

奇怪，我下载的数据集里2021_09_09_13_20_58是V2XSet内的。而且这个Scene因为camera 数据不全，所以我在这里 特地过滤了。

[Chinese-Coding]

对, 我也很疑惑这个问题. 我一开始以为是我OPV2V的数据集下载出现了错误, 然而我今天看到您新构建的数据集中也有这个文件, 所以感到困惑. 也许是数据集做了对应的更新, 不过移除之后就好了. 感谢您及时的回复. 🙏 🙏🙏

Yes, I am also confused about this problem. I thought at first that there was an error in downloading my OPV2V dataset, but today I saw that this file also exists in your newly constructed dataset, so I was confused. Maybe the dataset has been updated accordingly, but it will be fine after it is removed. Thank you for your timely reply. 🙏🙏🙏