Fine-Grained Recognition of Surface Targets with Limited Data

Runze Guo,Shaojing Su,Bei Sun,Xiaotian Qiu,Peng Wu,Zhen Zuo

doi:10.3390/electronics9122044

Abstract

Recognition of surface targets has a vital influence on the development of military and civilian applications such as maritime rescue patrols, illegal-vessel screening, and maritime operation monitoring. However, owing to the interference of visual similarity and environmental variations and the lack of high-quality datasets, accurate recognition of surface targets has always been a challenging task. In this paper, we introduce a multi-attention residual model based on deep learning methods, in which channel and spatial attention modules are applied for feature fusion. In addition, we use transfer learning to improve the feature expression capabilities of the model under conditions of limited data. A function based on metric learning is adopted to increase the distance between different classes. Finally, a dataset with eight types of surface targets is established. Comparative experiments on our self-built dataset show that the proposed method focuses more on discriminative regions, avoiding problems like gradient disappearance, and achieves better classification results than B-CNN, RA-CNN, MAMC, and MA-CNN, DFL-CNN.

Highlights

The recognition of surface targets has attracted broad interest from researchers in recent decades because of its diverse applications in marine monitoring and investigations, such as maritime patrols, rescue operations, and illegal-vessel screening, which are crucial for coastal countries [1]
There are a few works aimed at visible light images of surface targets
Learning are misidentified as buoys, and 6 aircrafts are misidentified as fishing boat in our model, while 24 out tocarriers the limited of self-built target datasets, this study uses transfer learning ensure of 100Owing aircraft are number misidentified as buoys, and 12 aircrafts are misidentified as fishingtoboat in that the network has sufficient feature expression capabilities

Summary

Introduction

The recognition of surface targets has attracted broad interest from researchers in recent decades because of its diverse applications in marine monitoring and investigations, such as maritime patrols, rescue operations, and illegal-vessel screening, which are crucial for coastal countries [1]. In the past two decades, various algorithm for surface target recognition have been proposed These methods can be roughly divided into moment-based [2], knowledge-based [3], model-based [4], and neural network-based [5,6]. These methods are mainly aimed at fixed scenes, static targets, and require a large amount of labeled data, which are difficult to implement in water scenes. A majority of existing methods used in detection and recognition are based on remote sensing images [7] rather than visible light images With this regard, there are a few works aimed at visible light images of surface targets

Methods

Results

Conclusion