UAV Detection Research Articles

Low-Latency Aerial Images Object Detection for UAV

Visual-based object detection has large applications in security and surveillance for unmanned aerial vehicles (UAVs). Meanwhile, the object detectors are required of low-latency and easy to be deployed on embedded onboard platforms. Aiming to address these problems, we present a PA-YOLOv3 aerial images object detector based on YOLOv3 and PANet algorithms, which can be deployed on embedded platforms. The PA-YOLOv3 model uses the dual-tower structure to improve the feature extraction and expression capabilities in feature fusion stage of the network. Besides, we propose a balanced pruning method to reduce the model size and the imbalance of different feature layers during pruning. After balanced pruning, the latency and size of the model are significantly decreased. Finally, we deploy and quantify the model on the embedded platform with TensorRT technology and apply the model on the UAV system for testing. The comprehensive experiments are executed on VisDrone2018 dataset and real-world scenarios. The experimental results show the inference speed of PA-YOLOv3 boost of about [Formula: see text] model pruning and quantization, while maintaining high detection accuracy.

Experimental Evaluation of Computer Vision and Machine Learning-Based UAV Detection and Ranging

We consider the problem of vision-based detection and ranging of a target UAV using the video feed from a monocular camera onboard a pursuer UAV. Our previously published work in this area employed a cascade classifier algorithm to locate the target UAV, which was found to perform poorly in complex background scenes. We thus study the replacement of the cascade classifier algorithm with newer machine learning-based object detection algorithms. Five candidate algorithms are implemented and quantitatively tested in terms of their efficiency (measured as frames per second processing rate), accuracy (measured as the root mean squared error between ground truth and detected location), and consistency (measured as mean average precision) in a variety of flight patterns, backgrounds, and test conditions. Assigning relative weights of 20%, 40% and 40% to these three criteria, we find that when flying over a white background, the top three performers are YOLO v2 (76.73 out of 100), Faster RCNN v2 (63.65 out of 100), and Tiny YOLO (59.50 out of 100), while over a realistic background, the top three performers are Faster RCNN v2 (54.35 out of 100, SSD MobileNet v1 (51.68 out of 100) and SSD Inception v2 (50.72 out of 100), leading us to recommend Faster RCNN v2 as the recommended solution. We then provide a roadmap for further work in integrating the object detector into our vision-based UAV tracking system.

Analysis of Micro-Doppler Signatures of Small UAVs Based on Doppler Spectrum

Most of the investigations on the micro-Doppler (MD) effect caused by a small unmanned aerial vehicle (UAV) have been conducted using joint time–frequency (JTF) images rather than the Doppler spectrum. On the other hand, several researchers utilized the Doppler spectrum instead of JTF images to observe the MD signature of a small UAV, and found the relationship between the spectral distribution of a small UAV and its physical specifications. However, the studies using the Doppler spectrum still lack concrete and theoretical foundations of the MD effects of a small UAV, focusing mainly on phenomena identified by measurement data. In this article, we establish the theoretical foundation connecting the MD signatures and motion dynamics of small UAVs based on the Doppler spectrum, and analyze their spectral distribution using simulations and measured data. In addition, experimental analysis is conducted using the data measured from various types of small UAVs considering the translational motion and aspect change. In contrast to already existing investigations, we completely explain and predict the changes on the Doppler spectrum relative to the physical specifications of a small UAV (e.g., blade length and rotor rotation rate). In particular, we show that the Doppler spectrum, compared to the JTF images, is a considerably simple and useful tool for analyzing the MD effects of small flying UAVs. The analysis results reveal that the MD features obtained from the measured echoes of small UAVs have considerable potential for detection and classification of small UAVs.

Landing Site Detection for UAVs Based on CNNs Classification and Optical Flow from Monocular Camera Images

The increased use of UAVs (Unmanned Aerial Vehicles) has heightened demands for an automated landing system intended for a variety of tasks and emergency landings. A key challenge of this system is finding a safe landing site in an unknown environment using on-board sensors. This paper proposes a method to generate a heat map for safety evaluation using images from a single on-board camera. The proposed method consists of the classification of ground surface by CNNs (Convolutional Neural Networks) and the estimation of surface flatness from optical flow. We present the results of applying this method to a video obtained from an on-board camera and discuss ways of improving the method.

UAV Collision Detection Algorithm based on Formulafor Fast Calculation

Collision Avoidance System (CAS) is known as a pre-collision system or a forward collision warning system, and research has first begun as a vehicle safety system. In this paper, we propose an algorithm for collision detection of UAV. The proposed algorithm uses a mathematical method and detects the collision of the UAV by modeling it in a two-dimensional plane. Using the mathematical modeling method, it is possible to determine the collision location of the UAV in advance. Experiments were conducted to measure the performance and accuracy of the proposed algorithm. In the experiment, we proceeded assuming three environments and were able to detect an accurate collision when the UAV moved. By applying the algorithm proposed in the paper to CAS, many collision accidents can be prevented. The proposed algorithm detects collisions through mathematical calculations. In addition, the movement time of the UAV was modeled in a 2D environment to shorten the calculation time.

Toward More Robust and Real-Time Unmanned Aerial Vehicle Detection and Tracking via Cross-Scale Feature Aggregation Based on the Center Keypoint

Unmanned aerial vehicles (UAVs) play an essential role in various applications, such as transportation and intelligent environmental sensing. However, due to camera motion and complex environments, it can be difficult to recognize the UAV from its surroundings thus, traditional methods often miss detection of UAVs and generate false alarms. To address these issues, we propose a novel method for detecting and tracking UAVs. First, a cross-scale feature aggregation CenterNet (CFACN) is constructed to recognize the UAVs. CFACN is a free anchor-based center point estimation method that can effectively decrease the false alarm rate, the misdetection of small targets, and computational complexity. Secondly, the region of interest-scale-crop-resize (RSCR) method is utilized to merge CFACN and region-of-interest (ROI) CFACN (ROI-CFACN) further, in order to improve the accuracy at a lower computational cost. Finally, the Kalman filter is adopted to track the UAV. The effectiveness of our method is validated using a collected UAV dataset. The experimental results demonstrate that our methods can achieve higher accuracy with lower computational cost, being superior to BiFPN, CenterNet, YoLo, and their variants on the same dataset.

Electrostatic Signal analysis of rotor UAV

With the rapid growth of unmanned aerial vehicle(UAV) market, incidents of small UAV flying without license or without permission frequently occur, which bring serious hidden dangers to the security of civil airports, sensitive targets and major activities. Rapid and accurate detection and recognition of small UAVs is the basis and premise of countermeasures against small UAVs. However, due to the fact that the production materials of small UAVs are mostly composite materials with good wave-absorbing and wave-penetrating properties and small size, the existing detection technologies such as radar and laser cannot effectively detect the signals of UAVs, so the detection accuracy is very low and the amount of information that can be obtained is small. In this paper, the most typical type of small UAV, rotor UAV, is taken as the research object to study the electrostatic signal of rotor UAV, which provides a theoretical basis for the electrostatic detection of rotor UAV.

Synthesis of algorithm of unmanned aerial vehicle detection by acoustic noise

The algorithm of detection of acoustic noise provided by an unmanned aerial vehicle (UAV) in the noise background due to wind is synthesized in the article. Creation of the algorithm has been carried out using the Neyman – Pearson lemma. The algorithm assumes a combination of the stages of wind noise coherent compensation and coherent accumulation of UAV’s acoustic noise sound pressure impulses. The coherent accumulation time matches doubled time of fluctuation correlation resulted by experimental research of acoustic noise of different types of UAVs. Efficiency of the developed algorithm of UAV detection depends on flight velocity, foreshortening, amount of blades and rotor turnovers of UAV as well as weather conditions. For the probability of a false alarm value of 10–4, the probability of correct UAV detection value of 0.9 is provided wherein signal-to-noise ratio has a value of 8 dB. These indicators correspond the detection range of 200 to 300 meters. The obtained results allow discussions about perspective of acoustic UAVs detection systems adaptation.

A Fast and Accurate Few-Shot Detector for Objects with Fewer Pixels in Drone Image

Unmanned aerial vehicles (UAVs) are important in modern war, and object detection performance influences the development of related intelligent drone application. At present, the target categories of UAV detection tasks are diversified. However, the lack of training samples of novel categories will have a bad impact on the task. At the same time, many state-of-the-arts are not suitable for drone images due to the particularity of perspective and large number of small targets. In this paper, we design a fast few-shot detector for drone targets. It adopts the idea of anchor-free in fully convolutional one-stage object detection (FCOS), which leads to a more reasonable definition of positive and negative samples and faster speed, and introduces Siamese framework with more discriminative target model and attention mechanism to integrate similarity measures, which enables our model to match the objects of the same categories and distinguish the different class objects and background. We propose a matching score map to utilize the similarity information of attention feature map. Finally, through soft-NMS, the predicted detection bounding boxes for support category objects are generated. We construct a DAN dataset as a collection of DOTA and NWPU VHR-10. Compared with many state-of-the-arts on the DAN dataset, our model is proved to outperform them for few-shot detection tasks of drone images.

UAV Detection and Identification Based on WiFi Signal and RF Fingerprint

The security threats caused by the popularity of Unmanned Aerial Vehicles (UAVs) have received much attention. In this paper, a UAV detection and identification system based on WiFi signal and radio frequency (RF) fingerprint is proposed. The system firstly conducts UAV detection and after the UAV is detected, fractal dimension (FD), axially integrated bispectra (AIB), and square integrated bispectra (SIB) are extracted as UAV RF fingerprints due to their applicability and reliability. Furthermore, we propose weighted AIB and SIB fingerprints to identify UAVs. Since the high dimensionality of AIB and SIB features, the principal component analysis (PCA) algorithm is applied to reduce the dimensionality of these two features. Then the neighborhood component analysis (NCA) algorithm is utilized to weight the dimensions of the two features, AIB and SIB. The extracted UAV fingerprints are stored as training data and test data, respectively. Finally, machine learning algorithms are utilized to identify UAVs. The identification results are as follows: In the indoor scenario, the average identification accuracy of three features (FD, AIB, SIB) are 100%, 97.23%, and 96.11%, respectively. In outdoor scenario, the identification accuracy of three features are 100%, 95.00%, and 93.50%, respectively.

FastUAV-NET: A Multi-UAV Detection Algorithm for Embedded Platforms

In this paper, a real-time deep learning-based framework for detecting and tracking Unmanned Aerial Vehicles (UAVs) in video streams captured by a fixed-wing UAV is proposed. The proposed framework consists of two steps, namely intra-frame multi-UAV detection and the inter-frame multi-UAV tracking. In the detection step, a new multi-scale UAV detection Convolutional Neural Network (CNN) architecture based on a shallow version of You Only Look Once version 3 (YOLOv3-tiny) widened by Inception blocks is designed to extract local and global features from input video streams. Here, the widened multi-UAV detection network architecture is termed as FastUAV-NET and aims to improve UAV detection accuracy while preserving computing time of one-step deep detection algorithms in the context of UAV-UAV tracking. To detect UAVs, the FastUAV-NET architecture uses five inception units and adopts a feature pyramid network to detect UAVs. To obtain a high frame rate, the proposed method is applied to every nth frame and then the detected UAVs are tracked in intermediate frames using scalable Kernel Correlation Filter algorithm. The results on the generated UAV-UAV dataset illustrate that the proposed framework obtains 0.7916 average precision with 29 FPS performance on Jetson-TX2. The results imply that the widening of CNN network is a much more effective way than increasing the depth of CNN and leading to a good trade-off between accurate detection and real-time performance. The FastUAV-NET model will be publicly available to the research community to further advance multi-UAV-UAV detection algorithms.

Unmanned Aerial Vehicle Operating Mode Classification Using Deep Residual Learning Feature Extraction

Unmanned Aerial Vehicles (UAVs) undoubtedly pose many security challenges. We need only look to the December 2018 Gatwick Airport incident for an example of the disruption UAVs can cause. In total, 1000 flights were grounded for 36 h over the Christmas period which was estimated to cost over 50 million pounds. In this paper, we introduce a novel approach which considers UAV detection as an imagery classification problem. We consider signal representations Power Spectral Density (PSD); Spectrogram, Histogram and raw IQ constellation as graphical images presented to a deep Convolution Neural Network (CNN) ResNet50 for feature extraction. Pre-trained on ImageNet, transfer learning is utilised to mitigate the requirement for a large signal dataset. We evaluate performance through machine learning classifier Logistic Regression. Three popular UAVs are classified in different modes; switched on; hovering; flying; flying with video; and no UAV present, creating a total of 10 classes. Our results, validated with 5-fold cross validation and an independent dataset, show PSD representation to produce over 91% accuracy for 10 classifications. Our paper treats UAV detection as an imagery classification problem by presenting signal representations as images to a ResNet50, utilising the benefits of transfer learning and outperforming previous work in the field.

RF-Based UAV Detection and Identification Using Hierarchical Learning Approach.

Unmanned Aerial Vehicles (UAVs) are widely available in the current market to be used either for recreation as a hobby or to serve specific industrial requirements, such as agriculture and construction. However, illegitimate and criminal usage of UAVs is also on the rise which introduces their effective identification and detection as a research challenge. This paper proposes a novel machine learning-based for efficient identification and detection of UAVs. Specifically, an improved UAV identification and detection approach is presented using an ensemble learning based on the hierarchical concept, along with pre-processing and feature extraction stages for the Radio Frequency (RF) data. Filtering is applied on the RF signals in the detection approach to improve the output. This approach consists of four classifiers and they are working in a hierarchical way. The sample will pass the first classifier to check the availability of the UAV, and then it will specify the type of the detected UAV using the second classifier. The last two classifiers will handle the sample that is related to Bebop and AR to specify their mode. Evaluation of the proposed approach with publicly available dataset demonstrates better efficiency compared to existing detection systems in the literature. It has the ability to investigate whether a UAV is flying within the area or not, and it can directly identify the type of UAV and then the flight mode of the detected UAV with accuracy around 99%.

Analysis Of Methods And Systems For Detecting And Countering Uavs

This article discusses methods for detecting UAVs in the radio frequency range. The analysis of the existing UAV detection systems is carried out. An overview of ways to counter UAVs is presented, in particular, by carrying out attacks on navigation and communication systems. A prototype of the UAV detection system, which is based on the UAV, is presented. The UAV detection function has been implemented based on the analysis of the radio frequency spectrum.

Comparative Monte Carlo Analysis of Background Estimation Algorithms for Unmanned Aerial Vehicle Detection

Background estimation algorithms are important in UAV (Unmanned Aerial Vehicle) vision tracking systems. Incorrect selection of an algorithm and its parameters leads to false detections that must be filtered by the tracking algorithm of objects, even if there is only one UAV within the visibility range. This paper shows that, with the use of genetic optimization, it is possible to select an algorithm and its parameters automatically. Background estimation algorithms (CNT (CouNT), GMG (Godbehere-Matsukawa-Goldberg), GSOC (Google Summer of Code 2017), MOG (Mixture of Gaussian), KNN (K–Nearest Neighbor–based Background/Foreground Segmentation Algorithm), MOG2 (Mixture of Gaussian version 2), and MEDIAN) and the reference algorithm of thresholding were tested. Monte Carlo studies were carried out showing the advantages of the MOG2 algorithm for UAV detection. An empirical sensitivity analysis was presented that rejected the MEDIAN algorithm.

UAV detection via long-time coherent integration for passive bistatic radar

Detection of unmanned aerial vehicle (UAV) is a challenging problem for passive bistatic radar (PBR) because of its low radar cross-section (RCS). Range migration (RM) will occur within one long coherent processing interval, which makes it difficult to increase coherent integration gain and improve radar detection ability. In order to eliminate the RM, a novel coherent integration method based on stretch processing and chirp-z transform (CZT) in the frequency domain is proposed. Firstly, stretch processing is applied to correct RM. Then, CZT is performed to mitigate the residual range offset and achieve coherent integration of the target's energy according to each Doppler cell. The advantage of the proposed method is that it can obtain superior coherent integration and detection performance compared to existing methods. Numerical and experimental results are presented to demonstrate the effectiveness of the proposed method.

RTT-Based Rogue UAV Detection in IoV Networks

Unmanned aerial vehicles (UAVs) are being used in different emerging domains for accomplishing many critical tasks. However, due to the various constraints, such as battery life, computational resources, etc., a UAV under a mission (M-UAV) often needs assistance from an edge/cloud server that is reachable from the M-UAV’s location. A connection between an M-UAV and edge server can be established via an access point or AP. Therefore, before sharing any sensitive information with the edge server, it is essential for an M-UAV to determine the legitimacy of the selected AP. Recently, some works in this direction indicate that a rogue UAV (R-UAV) can successfully mimic a legitimate AP for intercepting the communication channel. Hence, there should be a robust detection mechanism in place for addressing such a threat scenario. In this article, considering one of the emerging domains—the Internet of Vehicle (IoV) networks, at first, we show that communication in the IoV networks can get benefit from the presence of M-UAVs. However, as the link between the M-UAV and edge server can be intercepted by an R-UAV, the adversary may access the sensitive information from the IoV networks. Followed by this, we propose a <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">timing-based</i> algorithm for identifying the presence of rogue APs (or R-UAVs) in the channel. The M-UAV executes the timing-based algorithm, and the detection method <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">does not</i> require any auxiliary hardware or any modification to the network protocols for meeting the objective. Supported by an extensive evaluation study, we show that without any rigid restriction on the M-UAV’s speed (e.g., by limiting it to almost static) the proposed approach significantly enhances the detection accuracy (at least by a margin of 29.7% and 16.65%) compared to the state-of-the-art methods.

Air-to-Air Visual Detection of Micro-UAVs: An Experimental Evaluation of Deep Learning

This letter studies the problem of air-to-air visual detection of micro unmanned aerial vehicles (UAVs) by monocular cameras. This problem is important for many applications such as vision-based swarming of UAVs, malicious UAV detection, and see-and-avoid systems for UAVs. Although deep learning methods have exhibited superior performance in many object detection tasks, their potential for UAV detection has not been well explored. As the first main contribution of this letter, we present a new dataset, named Det-Fly, which consists of more than 13 000 images of a flying target UAV acquired by another flying UAV. Compared to the existing datasets, the proposed one is more comprehensive in the sense that it covers a wide range of practical scenarios with different background scenes, viewing angles, relative distance, flying altitude, and lightning conditions. The second main contribution of this letter is to present an experimental evaluation of eight representative deep-learning algorithms based on the proposed dataset. To the best of our knowledge, this is the first comprehensive experimental evaluation of deep learning algorithms for the task of visual UAV detection so far. The evaluation results highlight some key challenges in the problem of air-to-air UAV detection and suggest potential ways to develop new algorithms in the future. The dataset is available at https://github.com/Jake-WU/Det-Fly.

UAVData: A dataset for unmanned aerial vehicle detection

The unmanned aerial vehicles (UAVs) significantly contribute to the convenience and intelligence of life. However, the large use of UAVs also leads to high security risk. Only detecting the small and flying UAVs can prevent the safety accidents. UAV detection task could be regarded as a branch of object detection in flied of image processing. The advanced object detection models are mainly data driven, which depend on large-scale databases. The well-labeled datasets have proved to be of profound value for the effectiveness and accuracy in various object detection tasks. Thus, the first step of detecting UAVs is to build up a dataset of UAVs. In this study, we collect and release a dataset for UAV detection, called UAVData. To maintain the universality and robustness of the trained models, balloons and 6 types of UAVs are recorded in the dataset which totally consists of 13,803 well-labeled and recognizable images. We further conduct strong benchmarks using several advanced deep detection models, including faster R-CNN, SSD, YOLOv3. In addition, we utilize 4 different convolutional neural network models as the backbone models of these object detection methods to learn UAV-related features in images. By providing this dataset and baselines, we hope to gather researchers in both UAVs detection and machine learning field to advance toward the application.

A New Frequency Hopping Signal Detection of Civil UAV Based on Improved K-Means Clustering Algorithm

The rapid development of civil UAV promotes the social and economic development, and the frequent “flying illegally” events has brought great challenges to aviation safety and government supervision. The frequency hopping communication system used in UAV data transmission and control link has the advantages of anti-jamming and anti-interception, and its complex electromagnetic environment, which also brings great difficulties to UAV detection. In this paper, the detection of civil UAV is realized by frequency hopping signal monitoring. Firstly, by analyzing the signal characteristics of UAVs, an adaptive noise threshold calculation method is proposed for find the signals from spectrum data. Then, the improved clustering analysis algorithm is proposed based on constructed the waveform shape characteristics and peak characteristics of UAV frequency hopping signal. Finally, according to the designed experimental process, the experimental environment is set up, and the UAV monitoring, discovery and parameter estimation are realized by using the improved clustering analysis algorithm, and compared with K-means, K-means++, DBSCAN, Multi-hop and Auto-correlation methods. The results show that the method has certain robustness and has a good application prospect.