DPC: An amalgamated approach for drone package classification system

The logistics and e-commerce sectors have always incorporated parcel delivery and packing as their integral part. With the high amount of demand and supply in the domain of e-commerce, there is a high demand for drone-based delivery systems through which a certain amount of parcel packages can be delivered using drones. To assist the aforementioned domain, the proposed work aims to develop a drone package classification (DPC) system for the classification of packages according to their weight and the type of drone that can be employed to deliver the particular class of packages. The complete experiment has been executed using DL based hybrid approach consisting of convolutional neural network (CNN) and long short-term memory (LSTM) technique on a self-developed package5k dataset resulting in an overall accuracy of 93.82%. The entire investigation and suggested DL-based hybrid strategy have shown to be a successful, efficient, and sustainable method for the identification and classification of drone packages to create a successful and sustainable DPC system.