Feasibility Analysis of Optical UAV Detection Over Long Distances Using Robotic Telescopes

Ojdanic, Denis; Sinn, Andreas; Naverschnigg, Christopher; Schitter, Georg

doi:10.1109/taes.2023.3248560

Cited by 8 publications

(18 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The dataset used for fine-tuning contains 18000 images, with approximately two thirds being taken from [10] with additional images of UAVs being added from field tests using the presented telescope and camera system. The remaining 6000 images are taken from the Drone vs Bird (DvB) dataset [43].…”

Section: A Object Detectionmentioning

confidence: 99%

“…For the task of UAV detection different approaches exist including RADAR [6], [7], radio frequency [8], acoustic [9] and electro-optical detection [10]. Each of the mentioned methods have their benefits and drawbacks.…”

Section: Introduction 33mentioning

confidence: 99%

“…However, most of these systems ultimately rely on electro-optical sensors to perform object classification, as visual images can easily be interpreted by human operators or advanced computer vision algorithms. To extend the operational range of an electro-optical system, a narrow field of view (FoV) and a large optical aperture are necessary, which can be achieved by using telescopes [10]. To increase the situational awareness of such a system to a larger area, dedicated mounts enable pan and tilt motion [13].…”

Section: Introduction 33mentioning

confidence: 99%

See 2 more Smart Citations

Parallel Architecture for Low Latency UAV Detection and Tracking Using Robotic Telescopes

Ojdanić,

Naverschnigg,

Sinn

et al. 2024

IEEE Trans. Aerosp. Electron. Syst.

View full text Add to dashboard Cite

This paper presents the implementation of a multi-7 threaded parallel architecture, which enables telescope-based op-8 tical UAV detection and tracking in real-time. For efficient image 9 processing an accurate deep learning object detector is comple-10 mented in parallel by a fast object tracker. A transition strategy 11 between detector and tracker is introduced based on the tracker 12 reliability, which improves the object localization accuracy of 13 the system. The deep learning algorithm initializes the tracker 14 and in the subsequent frames the reliability of the tracker is 15 compared to the confidence value of each newly detected object to 16 determine whether a reinitialization is necessary. The implemented 17 architecture successfully demonstrates the parallel combination of 18 an FRCNN detector and a MEDIANFLOW tracker to achieve visual 19 UAV detection and tracking at 100 fps. The proposed reliability-20 based strategy outperforms a purely detector and tracker-based 21 strategy by 6% and 14% respectively in terms of intersection 22 over union at a threshold of 0.5, in scenarios, when the target 23 UAV is flying in front of a complex background. Additionally, the 24 implemented parallel architecture increases the probability for a 25 flight path estimation, which requires at least two localizations, by 26 49%, when compared to a non-parallel architecture. Field tests are 27 conducted with the proposed architecture using a telescope system 28 demonstrating UAV detection and tracking at 100 fps in distances 29 up to 4000 m in front of a clear background.30

show abstract

Section: A Object Detectionmentioning

confidence: 99%

Section: Introduction 33mentioning

confidence: 99%

Section: Introduction 33mentioning

confidence: 99%

See 1 more Smart Citation

Parallel Architecture for Low Latency UAV Detection and Tracking Using Robotic Telescopes

Ojdanić,

Naverschnigg,

Sinn

et al. 2024

IEEE Trans. Aerosp. Electron. Syst.

View full text Add to dashboard Cite

show abstract

“…The dataset used for training consists of 5000 images depicting different UAVs during daytime and clear weather conditions. About 50 % of the images are captured through various optical systems ranging from standard cameras to telescopes 9 and the remaining images are generated by blending cropped images of UAVs over different backgrounds. Therefore, the dataset contains images with various degrees of atmospheric blurring and backgrounds.…”

Section: Datasetmentioning

confidence: 99%

“…The detection of UAVs is an advanced research area with various approaches being investigated such as RADAR, 5 LiDAR, 6 radio frequency, 7 acoustics 8 and optics. 9 Most state of the art systems combine multiple of these technologies to benefit from the individual strengths forming a multispectral detection system. 10 Electrooptical systems are key components in all multispectral systems to ensure situational awareness, as captured images are easily interpreted.…”

Section: Introductionmentioning

confidence: 99%

Deep learning-based long-distance optical UAV detection: color versus grayscale

Ojdanic

Naverschnigg

Sinn

et al. 2023

Pattern Recognition and Tracking XXXIV

View full text Add to dashboard Cite

This paper presents a comparison between grayscale and color based deep learning algorithms for long distance optical UAV detection using robotic telescope systems. Three deep learning object detection algorithms are trained with a custom dataset consisting of RGB images and the performance is evaluated against the same algorithms trained with the same dataset converted to grayscale. Network training from scratch and fine-tuning are evaluated. The results for all algorithms show that fine-tuning with RGB images maximizes the detection performance and scores about 5 % better in terms of mean average precision (mAP(0.5)) compared to fine-tuning on grayscale images.

show abstract

Tiny drone object detection in videos guided by the bio-inspired magnocellular computation model

Wang,

Yang,

et al. 2024

Applied Soft Computing

View full text Add to dashboard Cite

Feasibility Analysis of Optical UAV Detection Over Long Distances Using Robotic Telescopes

Cited by 8 publications

References 30 publications

Parallel Architecture for Low Latency UAV Detection and Tracking Using Robotic Telescopes

Parallel Architecture for Low Latency UAV Detection and Tracking Using Robotic Telescopes

Deep learning-based long-distance optical UAV detection: color versus grayscale

Tiny drone object detection in videos guided by the bio-inspired magnocellular computation model

Contact Info

Product

Resources

About