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

Abstract

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.

Early Drought Detection in Maize Using UAV Images and YOLOv8+

“…The mAP values of Yolov8 are similar to those of Yolov7 in the initial stages, but as the epochs increase, its performance gradually surpasses that of Yolov7, stabilizing at around 68.4%. This suggests that Based on the results shown in Figure 13, which focuses on the evaluation metric mAP@50:95, the Yolov5s model exhibits relatively low mAP values in the initial epochs (0-50) [40]. However, as the number of epochs increases, the mAP value gradually improves and stabilizes around 66.8% at approximately epoch 200.…”

“…As an improved version of Yolov8, the Yolov8+ model maintains the highest mAP value (89.1%) throughout the entire training process, fully demonstrating its effectiveness in optimization strategy and structural design, thereby significantly enhancing model performance. Based on the results shown in Figure 13, which focuses on the evaluation metric mAP@50:95, the Yolov5s model exhibits relatively low mAP values in the initial epochs (0-50) [40]. However, as the number of epochs increases, the mAP value gradually improves and stabilizes around 66.8% at approximately epoch 200.…”

Malicious UAV Detection Over Rician Fading Channel: Performance Analysis

“…Recently, the aid of computer vision in target detection and tracking has developed rapidly due to its ability to provide high performance systems [30]- [32]. In [9], a Dalian University of Technology (DUT) Anti-UAV dataset was used on several existing detection algorithms for performance evaluation and comparison with a proposed detection and tracking algorithm that showed high performance by experiments.…”

A 3D Vision Cone Based Method for Collision Free Navigation of a Quadcopter UAV among Moving Obstacles

“…In [40], unlike the 3D occupancy map generation approaches for environment perception, this approach uses the object detection neural network to detect the drones. The authors tested a set of CNN-based object detection systems, such as Single Shot MultiBox Detector (SSD) with MobileNet v1 as backbone [41,42], Faster Region Based Convolutional Neural Networks (Faster-RCNN) [43], You Only Look Once v2 (YOLO v2) [44], and Tiny YOLO [45], to detect and track flying objects on the UAV's current flying trajectory.…”