This paper designs and implements a convolutional neural network for embedded systems, explores the deployment and realization method of a convolutional neural network in an embedded environment where both computational and storage resources are limited, and performs functional simulation verification of the realized network, aiming at solving the problem that the algorithmic complexity of the convolutional neural network has been increasing and the computational power of the embedded environment is insufficient. The experimental test results prove that this paper successfully deploys and realizes a convolutional neural network with a multilayer implicit layer under the embedded environment of scarce computational resources and limited storage resources, which can effectively complete the inference process of the network. The implementation method designed in this paper has a certain degree of flexibility, which lays a foundation for the construction of a reconfigurable intelligent computing platform that can deploy multiple convolutional neural networks under the embedded environment.