FeiGeChuanShu

Results 57 comments of FeiGeChuanShu

![QQ图片20210902085921](https://user-images.githubusercontent.com/80833738/131806701-d87e9832-9063-4caa-b836-918bf406c940.png) hi, maybe you should change the export code like this.

@caichuangcai 链接:https://pan.baidu.com/s/1oiLKiCuubAYrdEWIe1djRw 提取码:nvks

@sankexin param文件 加个参数5=1

> > param > > 在头部写进去报错,param里加哪里?怎么加的示例有吗? Reduction ReduceMean_611 1 1 src2_splitncnn_2 1088 0=3 1=0 -23303=1,1 4=1 5=1 类似这种 在后边加个5=1

> 你好,我使用YoloX官方的ONNX模型转成NCNN的文件,生成的param和bin文件大小完全一致,对比parame里面的内容也完全一致,替换后APP正常运行但是没有进行检测。请问这是什么问题呢? 非常感谢~ https://github.com/FeiGeChuanShu/ncnn-android-yolox/blob/main/app/src/main/jni/yolox.cpp#L352

已嫖 https://github.com/FeiGeChuanShu/ncnn_Android_face/tree/main/ncnn-android-yolov7_face

> Thanks for your nice work. > > But want to double confirm that if this is a complete demo... > > Look into source code, it only implemented TinyPose...

加载tflite权重,pytorch模仿原网络结构重新搭建网络,然后把加载的权重重新加载到网络中即可