说明：  ArmNN是Arm机构开源的基于arm嵌入式设备的inference框架，在Arm Cortex-A CPUs、Arm Mali GPUs、Arm Machine Learning processor都可以达到很高的加速效果。不过可惜的是，由于缺乏文档及教程，该框架在国内推广的程度不高，目前Github上star仅有359个。相对于其他竞品inference框架如NCNN、Tengine、Mace、MNN等等，ArmNN的知名度显得很小。不过笔者在项目的多次使用中，发现ArmNN是一个被低估的框架（在arm设备上的性能几乎都优于目前的前传框架），不愧是ARM家精心调教的SDK，对自家核心的性能调教到了极致。
(Armnn is an arm open source information framework based on ARM embedded devices. It can achieve high acceleration effect in arm cortex-a CPUs, arm Mali GPUs and arm machine learning processor. Unfortunately, due to the lack of documentation and tutorials, the framework is not widely promoted in China. At present, there are only 359 stars on GitHub. Compared with other competitive information frameworks, such as ncnn, Tengine, mace, MNN, etc., armnn has little popularity. However, the author found that armnn is an undervalued framework (the performance of arm devices is almost better than the current pre framework). It is worthy of being the arm family's carefully adjusted SDK, which is the ultimate performance tuning of its own core.)