ABSTRACT- The Vehicle detection is method to detect the vehicles in the image or video data. The vehicle detection is the branch of the object detection, where the vehicle is the primary object. The vehicle detection can be performed on various kinds of the vehicle data obtained from the horizontal, aerial, parking or road surveillance cameras. In this paper, the vehicle detection and classification method has been proposed by using the hybrid deep neural network over the image data and video obtained from the aerial and satellite images to determine the vehicle density. The non-negative matrix factorization (NMF) will be utilized for the feature extraction and compression for the purpose of vehicle detection and classification. The 2nd level feature compression will be performed to create the quick response vehicle detection and classification system. The model will be programmed to detect the maximum vehicles visible as full or partial object in the image. The vehicle density reporting, vehicle movement reporting and upside & downside reporting for highways will be performed to achieve the goal of the vehicle detection and classification. The aim of this review is to produce the robust algorithm to detect and analyze the vehicle features like whether the vehicle is heavy or light in the images and videos with higher accuracy and precision.