718c41634f
1.项目后端整体迁移至PaddleOCR-NCNN算法，已通过基本的兼容性测试 2.工程改为使用CMake组织，后续为了更好地兼容第三方库，不再提供QMake工程 3.重整权利声明文件，重整代码工程，确保最小化侵权风险 Log: 切换后端至PaddleOCR-NCNN，切换工程为CMake Change-Id: I4d5d2c5d37505a4a24b389b1a4c5d12f17bfa38c
10 lines
587 B
Markdown
10 lines
587 B
Markdown
## You can find keras2ncnn tools here
|
|
|
|
[https://github.com/MarsTechHAN/keras2ncnn](https://github.com/MarsTechHAN/keras2ncnn)
|
|
[https://github.com/azeme1/keras2ncnn](https://github.com/azeme1/keras2ncnn)
|
|
|
|
----
|
|
### From tensorflow 2.x, you can also export mlir and use mlir2ncnn which is maintained by the official.
|
|
|
|
The source code is located here: [https://github.com/Tencent/ncnn/tree/master/tools/mlir](https://github.com/Tencent/ncnn/tree/master/tools/mlir)
|
|
For Chinese, you can refer the guide here [https://zhuanlan.zhihu.com/p/152535430](https://zhuanlan.zhihu.com/p/152535430) |