UAV Target Tracking By Detection via Deep Neural Networks

Kassab, Mohamed; Maher, Ali; Elkazzaz, Fathy; Zhang, Baochang

doi:10.1109/icme.2019.00032

Cited by 16 publications

(8 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The stated work also performed real-flight validation tests and developed a novel camera handover strategy in order to enable long-term operation through several UAVs. As other close related works, some strategies use learning techniques for orientation tracking with respect to image patches based on state-of-the-art detectors and trackers [25,26].…”

Section: Related Workmentioning

confidence: 99%

Vision-Based Multirotor Following Using Synthetic Learning Techniques

Rodríguez-Ramos

Alvarez-Fernandez

Bavle

et al. 2019

Sensors

View full text Add to dashboard Cite

Deep- and reinforcement-learning techniques have increasingly required large sets of real data to achieve stable convergence and generalization, in the context of image-recognition, object-detection or motion-control strategies. On this subject, the research community lacks robust approaches to overcome unavailable real-world extensive data by means of realistic synthetic-information and domain-adaptation techniques. In this work, synthetic-learning strategies have been used for the vision-based autonomous following of a noncooperative multirotor. The complete maneuver was learned with synthetic images and high-dimensional low-level continuous robot states, with deep- and reinforcement-learning techniques for object detection and motion control, respectively. A novel motion-control strategy for object following is introduced where the camera gimbal movement is coupled with the multirotor motion during the multirotor following. Results confirm that our present framework can be used to deploy a vision-based task in real flight using synthetic data. It was extensively validated in both simulated and real-flight scenarios, providing proper results (following a multirotor up to 1.3 m/s in simulation and 0.3 m/s in real flights).

show abstract

Section: Related Workmentioning

confidence: 99%

Vision-Based Multirotor Following Using Synthetic Learning Techniques

Rodríguez-Ramos

Alvarez-Fernandez

Bavle

et al. 2019

Sensors

View full text Add to dashboard Cite

show abstract

“…Vision-based target tracking control methods can be divided into two categories: traditional target tracking methods [ 15 , 16 , 17 , 18 , 19 , 20 ] and learning-based target tracking methods [ 13 , 21 , 22 , 23 , 24 , 25 , 26 ]. The traditional vision-based UAV target tracking control scheme usually detects the target based on the color, shape and other characteristics, uses the vision-tracking algorithm to estimate the target state according to the image feature points, and then designs the corresponding control law to generate control instructions.…”

Section: Introductionmentioning

confidence: 99%

“…For the learning-based methods, the UAV target tracking control scheme inspired by the neural network takes the image as the input and directly outputs the action command through the neural network. Kassab et al [ 21 ] realized a target tracking system through two deep neural networks with the aid of image-based visual servo [ 22 ]. The proximity network estimates the relative distance between the UAV and the target based on the results of visual tracking, and the tracking network is used to control the relative azimuth between the UAV and the target.…”

Section: Introductionmentioning

confidence: 99%

Deep Reinforcement Learning-Based End-to-End Control for UAV Dynamic Target Tracking

et al. 2022

View full text Add to dashboard Cite

Uncertainty of target motion, limited perception ability of onboard cameras, and constrained control have brought new challenges to unmanned aerial vehicle (UAV) dynamic target tracking control. In virtue of the powerful fitting ability and learning ability of the neural network, this paper proposes a new deep reinforcement learning (DRL)-based end-to-end control method for UAV dynamic target tracking. Firstly, a DRL-based framework using onboard camera image is established, which simplifies the traditional modularization paradigm. Secondly, neural network architecture, reward functions, and soft actor-critic (SAC)-based speed command perception algorithm are designed to train the policy network. The output of the policy network is denormalized and directly used as speed control command, which realizes the UAV dynamic target tracking. Finally, the feasibility of the proposed end-to-end control method is demonstrated by numerical simulation. The results show that the proposed DRL-based framework is feasible to simplify the traditional modularization paradigm. The UAV can track the dynamic target with rapidly changing of speed and direction.

show abstract

“…Wenzhe Shi proposed an efficient sub-pixel CNN to improve the speed and quality of super-resolution reconstruction by introducing an effective sub-pixel convolutional layer and reducing the complexity of the system [15]. Mohamed A. Kassab used approaching and chasing networks to implement a real-time tracking control system based on full-vision depth object, which solves the problem of ambiguity between UAV yaw and lateral movement of moving objects [16]. Swathikiran Sudhakaran used LSTM with a pre-convolutional network to extract frame-level features, and proposed a simulated frame change model to improve the accuracy significantly [17].…”

Section: Introductionmentioning

confidence: 99%

An Improved Algorithm for Moving Object Detection in YOLO UAV Videos

Hu¹,

Hu²,

Wang³

et al. 2022

Journal of Compurters

View full text Add to dashboard Cite

<p>Recently, moving object detection (MOD) in UAV (Unmanned Aerial. Vehicle) videos has been widely used in many fields. However, different objects and different algorithms often result in different detection accuracy. SSD (Single Shot MultiBox Detector) series and YOLO (You Only Look Once) version 5 are two popular object detection model, and their performance are always evaluated and compared with other improved method for optimizing detection accuracy. In this paper, an improved YOLO_v5 detection algorithm was proposed to further improve the detection accuracy. It adopted a cascaded inter-frame verification mechanism which is based on the neural network and uses spatial information and integrates object speed and direction as well to improve the detection accuracy of moving objects. To evaluate its performance, the open UAV video data from Stanford University was used to test the algorithms, and three types of moving objects were analyzed. The experimental results demonstrate that the proposed MOD method can improve the detection accuracy of small moving objects, which have a good application value, and can lay a foundation for subsequent related studies.</p> <p> </p>

show abstract

UAV Target Tracking By Detection via Deep Neural Networks

Cited by 16 publications

References 11 publications

Vision-Based Multirotor Following Using Synthetic Learning Techniques

Vision-Based Multirotor Following Using Synthetic Learning Techniques

Deep Reinforcement Learning-Based End-to-End Control for UAV Dynamic Target Tracking

An Improved Algorithm for Moving Object Detection in YOLO UAV Videos

Contact Info

Product

Resources

About