Abstract:Precise aircraft recognition in low-resolution remote sensing imagery is a challenging yet crucial task in aviation, especially combat identification. This research addresses this problem with a novel, scalable, and AI-driven solution. The primary hurdle in combat identification in remote sensing imagery is the accurate recognition of Novel/Unknown types of aircraft in addition to Known types. Traditional methods, human expert-driven combat identification and image classification, fall short in identifying Novel classes. Our methodology employs similarity learning to discern features of a broad spectrum of military and civilian aircraft. It discerns both Known and Novel aircraft types, leveraging metric learning for the identification and supervised few-shot learning for aircraft type classification. To counter the challenge of limited low-resolution remote sensing data, we propose an end-to-end framework that adapts to the diverse and versatile process of military aircraft recognition by training a generalized embedder in fully supervised manner. Comparative analysis with earlier aircraft image classification methods shows that our approach is effective for aircraft image classification (F1-score Aircraft Type of 0.861) and pioneering for quantifying the identification of Novel types (F1-score Bipartitioning of 0.936). The proposed methodology effectively addresses inherent challenges in remote sensing data, thereby setting new standards in dataset quality. The research opens new avenues for domain experts and demonstrates unique capabilities in distinguishing various aircraft types, contributing to a more robust, domain-adapted potential for real-time aircraft recognition.

What problem does this paper attempt to address?

### Problems the Paper Aims to Solve This paper aims to address the challenge of accurately identifying known and unknown (novel) aircraft in low-resolution remote sensing images. Specifically, the paper focuses on issues in the aviation field, particularly in Combat Identification (CID). Traditional combat identification methods rely on human experts or image classification-based approaches, which perform poorly in identifying unknown types of aircraft. Therefore, this paper proposes a new, scalable, AI-driven solution that distinguishes various military and civilian aircraft, including known and unknown types, through similarity learning. ### Main Challenges 1. **Accurate Identification of Unknown Types of Aircraft**: Traditional methods have limitations in identifying newly emerging unknown aircraft types. 2. **Limited Data**: The amount of low-resolution remote sensing image data is limited, and the datasets contain noise and annotation errors. 3. **Dynamic Changes**: Aircraft designs are diverse and frequently change in appearance, increasing the difficulty of identification. 4. **Intra-class Variation**: Aircraft of the same category may exhibit different features under different conditions. ### Solution The paper proposes a method called "Intelligent Known and Novel Aircraft Recognition (INNAR)" to achieve its goals through the following steps: 1. **Embedding Space Construction**: Train a general embedder to generate vector representations that capture aircraft features. 2. **Few-shot Learning**: Utilize few-shot learning techniques to construct the embedding space with a small number of samples, addressing the data scarcity issue. 3. **Binary Classification Threshold Mechanism**: Introduce a threshold technique to determine whether an image belongs to a known category distribution, thereby distinguishing known and unknown aircraft. 4. **Final Classification**: For images belonging to known categories, use the K-nearest neighbors method for final classification. ### Contributions 1. **Automated Combat Identification**: Propose a new method for automated CID. 2. **Improved Dataset**: Conduct a thorough evaluation and cleaning of the MTARSI dataset, creating a higher quality dataset (MTARSI-INNAR). 3. **High-accuracy Identification**: Achieve high accuracy in identifying known and unknown aircraft categories. 4. **Method Validation**: Validate through experiments the inaccuracy of image classification methods on the original noisy dataset and propose a shift towards similarity and few-shot learning methods. ### Conclusion This study effectively addresses the challenge of identifying known and unknown aircraft in low-resolution remote sensing images by introducing the INNAR method, providing new benchmarks and solutions for the field of combat identification.

Intelligent Known and Novel Aircraft Recognition -- A Shift from Classification to Similarity Learning for Combat Identification

A Human–Computer Fusion Framework for Aircraft Recognition in Remote Sensing Images

Target detection and classification via EfficientDet and CNN over unmanned aerial vehicles

Development Of A Vision- based Anti-drone Identification Friend Or Foe Model To Recognize Birds And Drones Using Deep Learning

Detection and Recognition of Drones Based on a Deep Convolutional Neural Network Using Visible Imagery

DeepVision: Enhanced Drone Detection and Recognition in Visible Imagery through Deep Learning Networks

A Novel Method of Aircraft Detection under Complex Background Based on Circular Intensity Filter and Rotation Invariant Feature

MS-IAF: Multi-Scale Information Augmentation Framework for Aircraft Detection

Adaptive Component Discrimination Network for Airplane Detection in Remote Sensing Images

A Modified YOLOv5 Architecture for Aircraft Detection in Remote Sensing Images

A Transfer Learning Method for Aircrafts Recognition

Classi-Fly: Inferring Aircraft Categories from Open Data using Machine Learning

Light Weight Character and Shape Recognition for Autonomous Drones

A Deep Learning Approach for Drone Detection and Classification Using Radar and Camera Sensor Fusion

An Autonomous Self-Incremental Learning Approach for Detection of Cyber Attacks on Unmanned Aerial Vehicles (UAVs)

An improved deep learning method for flying object detection and recognition

Airplane Type Identification Based on Mask RCNN; An Approach to Reduce Airport Traffic Congestion

Intelligent Identification and Classification of Small UAV Remote Control Signals Based on Improved Yolov5-7.0

Combined Person Classification with Airborne Optical Sectioning

Aircraft Image Recognition Network Based on Hybrid Attention Mechanism.

ARNet: Prior Knowledge Reasoning Network for Aircraft Detection in Remote-Sensing Images