A Wireless Sensor Network (WSN) is incorporated with multi-capabilities to sense, calculation, data gathering, and communication. The network consists of small sensor nodes which are capable of monitoring and processing the data from a particular geographical position and transmits it to a remote location i.e. sink node or Base Station (BS). During the communication, extending the lifetime and stability of WSN remains challenging issue. Therefore, resource efficiency is a crucial factor in WSN to extend the network lifetime. In order to develop a Mean Shift Deming Regression-based Deep Multilayer Perceptive Neural Learning (MSDR-DMPNL) Model for dual cluster head selection to perform resource optimized data transmission in WSN. MSDR-DMPNL model comprised five layers, namely one input, three hidden layers, and one output layer for performing energy-efficient data transmission. The number of sensor nodes is considered as input in the input layer. After that, the energy and memory of the sensor node are computed in the hidden layer 1. Then, the information is transmitted to the hidden layer 2. In that layer, Mean Shift Node Clustering is carried out to perform the clustering process based on energy and memory. After that, the number of clusters is transmitted to the hidden layer 3 where Deming Regression is performed to select the dual cluster head (i.e., primary cluster head and secondary cluster head) in every cluster. Then the sensor node transmits the data packets to the primary cluster head through a neighboring node with high bandwidth availability. The primary cluster head transmits the collected data packets to the secondary cluster head. Finally, it transmits the received data packets to the base station. In this way, efficient resource-efficient data transmission is carried out. Simulation is conducted on factors such as clustering accuracy, energy consumption, packet delivery ratio, and delay with respect to a number of sensor nodes and packets.