For convolutional neural networks (CNNs) that have a large volume of input\ndata, memory management becomes a major concern. Memory cost reduction can be\nan effective way to deal with these problems that can be realized through\ndifferent techniques such as feature map pruning, input data splitting, etc.\nAmong various methods existing in this area of research, splitting the network\nstructure is an interesting research field, and there are a few works done in\nthis area. In this study, the problem of reducing memory utilization using\nnetwork structure splitting is addressed. A new technique is proposed to split\nthe network structure into small parts that consume lower memory than the\noriginal network. The split parts can be processed almost separately, which\nprovides an essential role for better memory management. The split approach has\nbeen tested on two well-known network structures of VGG16 and ResNet18 for the\nclassification of CIFAR10 images. Simulation results show that the splitting\nmethod reduces both the number of computational operations as well as the\namount of memory consumption.\n