SwakinX
SwakinX
我遇到问题测试了下，文字（宽或高）在图像中占比过高时就会这样，主要是v5模型会直接错误，不过v4也可能会出现重复识别的情况，将边缘扩展一下减少占比就能正常识别了，不能精确裁剪识别 用v5会错误 用v4会出现重复识别
@SWHL 确实可行，之前尝试单模块时是多行文本，效果并不好就没考虑了，谢谢提醒
https://github.com/Nuitka/Nuitka/issues/1699#issuecomment-1465204977 deleting openvino.dll from top level dist folder
> Unfortunately, I get an other error with/without the libopenvino.so.2510 file, because there are no "available devices": > > ``` > core = ov.Core() > print(f"Available devices: {core.available_devices}") > ```...
@f-domonkos It seems you misunderstood me. I'm not asking you to randomly delete other files. Instead, it asks you to move the entire openvino library file to the directory of...
我觉得可以再加一个全局配置`default_model_path`来自定义默认模型路径，这样没特殊情况就不用分别去修改模型路径，也能正常使用自动下载模型，同时添加一个方法允许在代码中直接设置默认模型路径