A Method for Special Vehicle Recognition Based on Deep-Transfer Model

摘要

As an application of image recognition, special vehicle recognition is very important in military field. This paper proposes a deep-transfer model (DTM) to overcome the problems in existing recognition methods. The DTM combines deep-learning and transfer-learning to solve the difficulty in training deep model with insufficient simples, improving the performance of the recognition algorithm. At last, the special vehicle dataset is built to evaluate the proposed DTM method. The results demonstrate that the DTM method outperforms the existing method in special vehicle recognition application.