夏至

只是使用中遇到的也不太清楚原理。。 感觉跟MybatisX插件有关的原因是打开插件飘红，关掉就好了。 还有`Alt+Enter`选择 Suppress for tag 之后会添加这个注释  

几万就是完全没有识别到= = 因为物品的数目是会滚动的，所以截取的图片的位置可能就不准确。稍微改了下逻辑，可以尝试下新版本好用了不。 尽量用 1600 * 900 的分辨率吧，在我本地测试是没有问题的。

I have no idea how to pass it

Maybe caused by the dependency modules, like moving the mouse cursor and capture the screen or something... But these modules are not directly developed by me. I develop the app...

The error doesn't appear to be caused by this application. The crash report shows the error was caused by `ucrtbase.dll`, and this dll comes with Windows. I googled it and...

OCR the small numbers is not quite stable. Try different resolutions.

I have never received isssues like this, can you provide more details?

``` ...... 'subattr_3': '生命值+1,005' ...... ``` 这里识别到的生命值+1,005是无效的数值（因为可能的数值的组合209/239/269/299不可能组合出1005），所以对识别结果做了过滤

不会提高。模型的训练基于随机生成的词条图片，而非实际的圣遗物截图。 > 这种识别错误导致圣遗物被过滤掉的情况，似乎不会生成截图。 考虑之后的版本进行优化，在发生这种错误时截图保存

As mentioned in [Readme](https://github.com/daydreaming666/Amenoma/blob/3.0/README_en.md), > Q: The numbers of items/materials are not completely accurate and some fail to be scanned. > A: This is normal because it scans the screen...