The dawn of AI in industries and spreading deep neural networks in all fields particularly in the field of computer vison make a significant need to use these kind of architect on everyday human being life. At the same time, huge complexity computation and not-cheap hardware such as GPUs are considered as the biggest obstacle to attain deep neural networks on edge devices such as Mobiles, simple boards and etc., which have numerous limitation on memory, computation and power. In this topic we are going to use novel AI technique in order to optimize powerful neural network on the edge devices whose numerous limitation in compared with typical GPUs. To illustrate, consider figure bellow that shows Nvidia Jetson Tx2 and its specific which is like typical cellphones, if this boards be able to run models which are real-time and high-performance, a significant step to will be taken to pacify artificial intelligence.