Potential Version Conflict Bugs Report

[WiGig11]

zh_CN:
使用本项目完成aachen的visual localization测试，记录全部过程：
1.搭建环境：
本人的环境是： 5.15.0-130-generic #140~20.04.1-Ubuntu SMP x86_64 x86_64 x86_64 GNU/Linux
按照官方的环境搭建方式遇到的错误是：虽然官方给出的版本是3.7即可，但是3.7是安装不上pycolmap的。
解决过程
1：升级到3.11：碰到no attribute error: issue49
2：降级到3.10：IndexError: _Map_base::at issue 116
最后的解决方式；安装0.6.1

2.修改路径：不赘述

3.netvlad非常慢，不是推理慢，经过断点调试pdb之后发现应该是权重下载很慢，所以根据issue 223,验证hub的对应权重（应该有500mb左右）

4.数据集：尽管在aachen和aachenv1.1的md中给出了指令，但是又有wget的问题，所以我强烈建议仔细阅读aachen和aachenv1.1的md文件，这样可以有效的帮助你理解两个数据集的差异和用法

EN：

Use this project to complete the visual localization test for Aachen and record the entire process:

1. Environment Setup:
   My environment is: 5.15.0-130-generic I got error. when install pycolmap #140~20.04.1-Ubuntu SMP x86_64 x86_64 x86_64 GNU/Linux
   The error encountered while following the official environment setup instructions was: Although the official documentation states that Python version 3.7 is sufficient, it is impossible to install pycolmap with Python 3.7.
   Resolution Process:

2. Upgrading to Python 3.11 resulted in an error: no attribute error. Issue #49

3. Downgrading to Python 3.10 caused another error: IndexError: _Map_base::at. Issue #116
   Final Solution: Install pycolmap version 0.6.1.

4. Path Modification: Details omitted.

5. NetVLAD Performance:
   NetVLAD is extremely slow, but not due to inference speed. After debugging with pdb, it was determined that the slowness likely stems from downloading weights. According to Issue #223, verify the corresponding weights on PyTorch Hub (the size should be around 500MB).

6. Dataset:
   Although the instructions for both Aachen and Aachen v1.1 datasets are provided in their respective .md files, there are issues with using wget. I strongly recommend carefully reading the .md files for both Aachen and Aachen v1.1 datasets. This will help you better understand the differences between the two datasets and how to use them effectively.

[WiGig11]

Also，you may find the issue:issue266 helpful. the author and the user had a discussion on the proper usage