Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Drone-type-Set: Drone types detection benchmark for drone detection and tracking

Kholoud AlDosari, AIbtisam Osman, Omar Elharrouss, Somaya AlMaadeed, Mohamed Zied Chaari

TL;DR

This paper addresses the rising security risks from unauthorized UAVs and the need to identify drone types using image-based detection. It introduces a Drone-type Dataset with bounding-box annotations for seven drone types collected from online sources and evaluates four detectors—YOLOv3, YOLOv4, YOLOv5, and Detectronv2—across multiple datasets. Key contributions include (i) the new drone-type dataset, (ii) a standardized training/evaluation protocol, and (iii) a comprehensive benchmark showing how model choice and dataset characteristics affect $mAP$ and related metrics. The findings reveal model- and data-dependent performance, with YOLOv3 attaining the highest $mAP$ on the proposed dataset and Detectronv2 excelling on some existing datasets, providing a valuable resource for security applications and further research.

Abstract

The Unmanned Aerial Vehicles (UAVs) market has been significantly growing and Considering the availability of drones at low-cost prices the possibility of misusing them, for illegal purposes such as drug trafficking, spying, and terrorist attacks posing high risks to national security, is rising. Therefore, detecting and tracking unauthorized drones to prevent future attacks that threaten lives, facilities, and security, become a necessity. Drone detection can be performed using different sensors, while image-based detection is one of them due to the development of artificial intelligence techniques. However, knowing unauthorized drone types is one of the challenges due to the lack of drone types datasets. For that, in this paper, we provide a dataset of various drones as well as a comparison of recognized object detection models on the proposed dataset including YOLO algorithms with their different versions, like, v3, v4, and v5 along with the Detectronv2. The experimental results of different models are provided along with a description of each method. The collected dataset can be found in https://drive.google.com/drive/folders/1EPOpqlF4vG7hp4MYnfAecVOsdQ2JwBEd?usp=share_link

Drone-type-Set: Drone types detection benchmark for drone detection and tracking

TL;DR

This paper addresses the rising security risks from unauthorized UAVs and the need to identify drone types using image-based detection. It introduces a Drone-type Dataset with bounding-box annotations for seven drone types collected from online sources and evaluates four detectors—YOLOv3, YOLOv4, YOLOv5, and Detectronv2—across multiple datasets. Key contributions include (i) the new drone-type dataset, (ii) a standardized training/evaluation protocol, and (iii) a comprehensive benchmark showing how model choice and dataset characteristics affect and related metrics. The findings reveal model- and data-dependent performance, with YOLOv3 attaining the highest on the proposed dataset and Detectronv2 excelling on some existing datasets, providing a valuable resource for security applications and further research.

Abstract

The Unmanned Aerial Vehicles (UAVs) market has been significantly growing and Considering the availability of drones at low-cost prices the possibility of misusing them, for illegal purposes such as drug trafficking, spying, and terrorist attacks posing high risks to national security, is rising. Therefore, detecting and tracking unauthorized drones to prevent future attacks that threaten lives, facilities, and security, become a necessity. Drone detection can be performed using different sensors, while image-based detection is one of them due to the development of artificial intelligence techniques. However, knowing unauthorized drone types is one of the challenges due to the lack of drone types datasets. For that, in this paper, we provide a dataset of various drones as well as a comparison of recognized object detection models on the proposed dataset including YOLO algorithms with their different versions, like, v3, v4, and v5 along with the Detectronv2. The experimental results of different models are provided along with a description of each method. The collected dataset can be found in https://drive.google.com/drive/folders/1EPOpqlF4vG7hp4MYnfAecVOsdQ2JwBEd?usp=share_link
Paper Structure (11 sections, 2 equations, 8 figures, 3 tables)

This paper contains 11 sections, 2 equations, 8 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Related works
  3. Proposed method
  4. Drone-type Dataset
  5. Drone detection and recognition models
  6. Experimental results
  7. Dataset configuration
  8. Training setup
  9. Evaluation Metrics
  10. Evaluation and discussion
  11. Conclusion

Figures (8)

  • Figure 1: Drone-type dataset With annotation of different scale.
  • Figure 2: Training outputs of YOLOv5 on Dataset 2.
  • Figure 3: (a) YOLOv5: confusion matrix of D3-a. (b) YOLOv5: confusion matrix of D3-b
  • Figure 4: Precision-recall curve for propsoed Drone-Type dataset Using YOLOv5
  • Figure 5: Drone detection results on Dataset D1. (First row) YOLOv4. (Second row) YOlOv5. (Third row) Detectronv2.
  • ...and 3 more figures