Over the last few years, the accelerated development of e-commerce and logistics services made the quality of shipment more significant. During transportation, parcels are damaged and this results in loss of money, unsatisfied customers and inefficiencies in the operations. The process of the manual inspection of parcels is time-consuming, error prone and cannot be implemented in large-scale logistics settings. To solve these problems, the proposed project will suggest a deep-learning-based parcel-damage classification system based on the current abnormal event detection models. The Convolutional Neural Network (CNN) with transfer learning is employed to classify by itself (without human intervention) parcel images as damaged and non-damaged, and into various degrees of the severity of the damages. The results of the experiment indicate that the system is effective and can be implemented in the logistics processes of the real-time.
MAHESH et al. (Thu,) studied this question.