# paddle-2.3.2-cp37m-linux-aarch64.whl下载
dlib19.19.0m>cp m>m>37 m>m>cp m>m>37 m>m linux_aarch64.whl
下载本资源请参考博文：“RK3399安装python三方包face_recognition” 因为rk3399Pro平台编译时间
scipy1.1.0m>cp m>m>37 m>m>cp m>m>37 m>m linux_aarch64.whl
Scipy 1.1.0 for Ubuntu arm64 with Python 3.7. Suitable for Raspb
numpy1.18.5m>cp m>m>37 m>m>cp m>m>37 m>m linux_aarch64.whl
64位,arm处理器架构的numpy,采用交叉编译的方式生成,安装tensorflow的必备依赖,可以安装在树莓派4b上,适配3
grpcio1.32.0m>cp m>m>37 m>m>cp m>m>37 m>m linux_aarch64.whl
64位,arm处理器架构的grpcio,采用交叉编译的方式生成,可以安装在树莓派4b上,适配3.7版本的64位python
scipy1.4.1m>cp m>m>37 m>m>cp m>m>37 m>m linux_aarch64.whl
64位,arm处理器架构的scipy,采用交叉编译的方式生成,安装tensorflow的必备依赖,可以安装在树莓派4b上,适配3
numpy1.18.2cp37cp37mlinux_aarch64.whl
NumPy1.18.2forRaspberryPi3B&4BforPython3.7,withUbuntu19.10andlat
tensorflow1.15.0cp37cp37mlinux_aarch64.whl
Tensorflow1.15.0forUbuntuarm64withPython3.7.SuitableforRaspberry
scipy1.3.3cp37cp37mlinux_aarch64.whl
Scipy1.3.3forUbuntuarm64withPython3.7.SuitableforRaspberryPi3B/3
torch1.4.0a0加72e1771m>cp m>m>37 m>m>cp m>m>37 m>m linux_aarch64.whl
树莓派上编译的,不支持CUDA和MKLDNN的,CUDA是nv的,MKLDNN是intel的, 拿树莓派也只做推理,分布式也不支
tensorflow2.0.0cp37nonelinux_aarch64.whl
以python3.7为主体的tensorflow2.0的安装包。可以通过pip3加此安装包的路径进行安装