mirror of
https://github.com/babalae/better-genshin-impact.git
synced 2026-01-25 19:32:38 +08:00
5b3bac478d
* 更新多个NuGet包至最新版本 * 重构模型加载以适应yolosharp * feat: 改变接口。TensorRT缓存的初步支持，修改配置项。 * 更新依赖并调整变量顺序，修复加载问题 * 更新AvalonEdit和Microsoft.ML.OnnxRuntime包至最新版本,以修复问题 * fix: downgrade Microsoft.ML.OnnxRuntime.DirectML to version 1.21.0 * typo * fix: change log level from warning to error for ONNX provider loading failure * 增加 paddle ocr 的 onnx 模型 * feat: add PaddleOCR models for Chinese, English, and Latin recognition * 使用cv的DNN生成Tensor，加速Yap文字识别 * feat: 尝试搓一个onnx的ocr * clean up code * chore: update OpenCvSharp4 package versions to 4.10.0.20241108 * 修复因格式化代码而丢的引用 * chore: update Microsoft.ML.OnnxRuntime.DirectML package to version 1.21.1 and improve logging for ONNX provider initialization * chore: 等yolosharp更新再升级onnx * chore: add Microsoft.ML.OnnxRuntime.Managed package and clean up logging in Det class * fix: refactor output tensor handling in Det class for improved clarity * 补充注释，修复DML的OCR问题 * 默认OCR推理使用CPU,整理配置 * fix error NETSDK1152: 找到了多个具有相同相对路径的发布输出文件 * fix(logging): enhance debug log for ONNX initialization with provider details * 修复TensorRT模型缓存的加载问题 * fix(onnx): improve cached model retrieval and add file existence check * fix(ocr): replace SrcGreyMat with SrcMat for region of interest processing * fix(onnx): add file existence check for cached model and adjust session options for DirectML provider * 增加硬件加速配置UI界面 * 移除旧的OCR模型 * 错别字 --------- Co-authored-by: 辉鸭蛋 <huiyadanli@gmail.com>