With the faster development and advancement in surveillance and intelligence field such as the use of Unmanned Aerial Vehicles (UAVs) shows significant importance of object detection and movement understanding. However, data is gathered from varied locations, and the classification of that abundant amount of data is a challenging process. Thus, vehicle identification in aerial images is an essential and important research area. Therefore, a Selection Decision-Classification (􀜵􀝁􀝈􀜦􀝁􀝏- 􀜥􀝈􀜽􀝏) model in the Convolutional Neural Network (CNN) is adopted in this work to efficiently classify a large amount of vehicular data and correctly identify which data belongs to which class. The proposed 􀜵􀝁􀝈􀜦􀝁􀝏􀜥􀝈􀜽􀝏 modelis transformed into the Selection and Decision Network (􀜵􀝁􀝀􀝁􀜰􀝁􀝐) and Classification Network (􀜥􀝈􀝏􀜰􀝁􀝐). Here, the 􀜵􀝁􀝀􀝁􀜰􀝁􀝐 model works on the essential image region selection and generation of preliminary weights and the 􀜥􀝈􀝏􀜰􀝁􀝐 model performs an efficient training and performs classification process. The event Recognition Aerial (ERA) dataset is utilized to evaluate the classification performance of the proposed 􀜵􀝁􀝈􀜦􀝁􀝏􀜥􀝈􀜽􀝏 model. Binary classification is performed to test the model and its efficiency. The obtained classification accuracy using Traffic-Collision is 56.7%and Traffic-Congestion is 69.8%.