# An Efficient Detection Approach for Unmanned Aerial Vehicle (UAV) Small Targets Based on Group Convolution

^{1}

^{2}

^{3}

^{4}

^{5}

^{*}

## Abstract

**:**

## 1. Introduction

## 2. Related Work

## 3. Building a Drone Dataset

## 4. Improved Faster-RCNN Algorithm

#### 4.1. Types of Prior Bounding Boxes Using Cluster Analysis

_{1}and the predicted box is box

_{2}. The definition of the box value and the calculation formula of the IoU value are shown in the following:

_{1}and b

_{2}represent box

_{1}and box

_{2}.

_{1}, b

_{2}) is used, the larger the IoU (b

_{1}, b

_{2}) is, the higher the degree of coincidence is. We hope that the higher the degree of coincidence is, the shorter the distance is. The Kmeans clustering method can be better. Therefore, the IoU (b

_{1}, b

_{2}) cannot be used directly. The “1” should be added to the negative sign to ensure that the shorter the distance is, the better it is.

#### 4.2. Resnext50 Backbone

_{in}, n is convolution kernels, and g is the number of groups of group convolution. The principle of parameter optimization is as follows:

#### 4.3. Prior Bounding Box Design for the Receptive Field Area in FPN

^{2}or so, and the bottom layer of FPN has smaller target features. Thus, the minimum prior bounding box is designed to be the average size of the drone, to make full use of the low-level information to detect as many small drone targets as possible and calculate the size of all layers. After the receptive field, this paper set the prior bounding box size to four categories: 16

^{2}, 32

^{2}, 64

^{2}, and 128

^{2}to ensure that drones of each size can be selected. Since the overall size of the drone is between 10 × 8 and 65 × 21 during the period, the length–width ratio of the prior bounding box is designed with ratios of 2:1, 1:1, and 1:2 to ensure that the prior bounding box can match the appropriate target.

#### 4.4. Gse Attention Mechanism

## 5. Experimental Results

#### 5.1. Train the Network

#### 5.2. Result Analysis

## 6. Conclusions

## Author Contributions

## Funding

## Institutional Review Board Statement

## Informed Consent Statement

## Data Availability Statement

## Conflicts of Interest

## References

- Liu, Y.; Wan, X.; Tang, H.; Yi, J.; Cheng, Y.; Zhang, X. Digital Television Based Passive Bistatic Radar System for Drone Detection. In Proceedings of the 2017 IEEE RADAR Conference (RADARCONF), Seattle, WA, USA, 8–12 May 2017; pp. 1493–1497. [Google Scholar]
- Ren, S.; He, K.; Girshick, B.R.; Sun, J. Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks. IEEE Trans. Pattern Anal. Mach. Intell.
**2017**, 39, 1137–1149. [Google Scholar] [CrossRef] [PubMed][Green Version] - Dai, J.; Li, Y.; He, K.; Sun, J. R-FCN: Object Detection via Region-based Fully Convolutional Networks. In Advances in Neural Information Processing Systems 29 (NIPS 2016) Barcelona, Spain, 5–10 December 2016; NIPS: Barcelona, Spain, 2016. [Google Scholar]
- Redmon, J.; Farhadi, A. YOLOv3: An Incremental Improvement. arXiv
**2018**, arXiv:1804.02767v1. [Google Scholar] - Wang, C.Y.; Bochkovskiy, A.; Liao, M.H.Y. Scaled-YOLOv4—Scaling Cross Stage Partial Network. In Proceedings of the CVPR 2021, Nashville, TN, USA, 19–25 June 2021; pp. 13029–13038. [Google Scholar]
- Etten, V.A. You Only Look Twice: Rapid Multi-Scale Object Detection In Satellite Imagery. arXiv
**2018**, arXiv:1805.09512. [Google Scholar] - Li, X.; Wang, W.; Hu, X.; Li, J.; Tang, J.; Yang, J. Generalized Focal Loss V2—Learning Reliable Localization Quality Estimation for Dense Object Detection. In Proceedings of the CVPR 2021, Nashville, TN, USA, 19–25 June 2021; pp. 11632–11641. [Google Scholar]
- Ashraf, W.M.; Sultani, W.; Shah, M. Dogfight—Detecting Drones from Drones Videos. In Proceedings of the CVPR 2021, Nashville, TN, USA, 19–25 June 2021; pp. 7067–7076. [Google Scholar]
- Zou, Z.; Shi, Z.; Guo, Y.; Ye, J. Object Detection in 20 Years: A Survey. arXiv
**2019**, arXiv:1905.05055. [Google Scholar] - Liu, L.; Ouyang, W.; Wang, X.; Fieguth, W.P.; Chen, J.; Liu, X.; Pietikainen, M. Deep Learning for Generic Object Detection: A Survey. Int. J. Comput. Vis.
**2020**, 128, 261–318. [Google Scholar] [CrossRef][Green Version] - Xinbo, G.; Mengcheng, M.; Haitao, W.; Jiaxu, L. Research progress of small target detection processing. J. Data Acquis. Process.
**2021**, 36, 391–417. [Google Scholar] [CrossRef] - Cai, Z.; Vasconcelos, N. Cascade R-CNN: High Quality Object Detection and Instance Segmentation. IEEE Trans. Pattern Anal. Mach. Intell.
**2021**, 43, 1483–1498. [Google Scholar] [CrossRef] [PubMed][Green Version] - He, K.; Gkioxari, G.; Doll‘ar, P.; Girshick, B.R. Mask R-CNN. In Proceedings of the ICCV 2020, Venice, Italy, 14–19 June 2020; pp. 386–397. [Google Scholar]
- Yu, X.; Gong, Y.; Jiang, N.; Ye, Q.; Han, Z. Scale Match for Tiny Person Detection. In Proceedings of the WACV 2020, Snowmass, CO, USA, 1–5 March 2020; pp. 1246–1254. [Google Scholar]
- Kisantal, M.; Wojna, Z.; Murawski, J.; Naruniec, J.; Cho, K. Augmentation for small object detection. In Proceedings of the Computer Vision and Pattern Recognition, Long Beach, CA, USA, 16–20 June 2019. [Google Scholar]
- Xie, S.; Girshick, B.R.; Dollár, P.; Tu, Z.; He, K. Aggregated Residual Transformations for Deep Neural Networks. In Proceedings of the 30th IEEE Conference on Computer Vision and Pattern Recognition (CVPR 2017), Honolulu, HI, USA, 21–26 July 2017; pp. 5987–5995. [Google Scholar]
- He, K.; Zhang, X.; Ren, S.; Sun, J. Deep Residual Learning for Image Recognition. In Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR2016), Las Vegas, NV, USA, 27–30 June 2016; pp. 770–778. [Google Scholar]
- PengFei, L.; Hai, Z.; Shuichun, F.; Chunjiang, B. Multi scale low altitude UAV detection based on improved SSD Computer engineering and design. Comput. Eng. Des.
**2021**, 42, 3277–3285. [Google Scholar] [CrossRef] - Lin, T.Y.; Doll’ar, P.; Girshick, B.R.; He, K.; Hariharan, B.; Belongie, J.S. Feature Pyramid Networks for Object Detection. In Proceedings of the 30th IEEE Conference on Computer Vision and Pattern Recognition (CVPR 2017), Honolulu, HI, USA, 21–26 July 2017; pp. 936–944. [Google Scholar]
- Hu, J.; Shen, L.; Sun, G. Squeeze-and-Excitation Networks. In Proceedings of the CVPR 2019, Long Beach, CA, USA, 16–20 June 2019. [Google Scholar]

**Figure 7.**Comparison chart of experimental results. (

**a**) Faster-rcnn; (

**b**) GC-Faster-rcnn. The boxes in these figures are the detection accuracy.

Backbone | Layers | Parameters | Size (M) |
---|---|---|---|

Resnet50 | 3 4 6 3 | 25,557,032 | 97.49 |

Resnext50 | 3 4 6 3 | 25,028,904 | 95.48 |

Resnext50 | 3 4 6 2 | 20,531,496 | 78.32 |

Algorithm | Backbone (50 Layers) | mAP0.5/% | mAP0.75/% |
---|---|---|---|

Faster-rcnn | Resnet | 71.5 | 27.6 |

Resnet + Y | 89.8 | 38.2 | |

Resnet + Y + Gse | 93.2 | 44.0 | |

GC-Faster-rcnn | Resnext + Y | 91.3 | 38.2 |

Resnext + Y + Gse | 94.8 | 41.9 |

Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. |

© 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).

## Share and Cite

**MDPI and ACS Style**

Cheng, J.; Liu, Y.; Li, G.; Li, J.; Peng, J.; Hong, J.
An Efficient Detection Approach for Unmanned Aerial Vehicle (UAV) Small Targets Based on Group Convolution. *Appl. Sci.* **2022**, *12*, 5402.
https://doi.org/10.3390/app12115402

**AMA Style**

Cheng J, Liu Y, Li G, Li J, Peng J, Hong J.
An Efficient Detection Approach for Unmanned Aerial Vehicle (UAV) Small Targets Based on Group Convolution. *Applied Sciences*. 2022; 12(11):5402.
https://doi.org/10.3390/app12115402

**Chicago/Turabian Style**

Cheng, Jianghao, Yanyan Liu, Guoning Li, Jin Li, Jiantao Peng, and Jintao Hong.
2022. "An Efficient Detection Approach for Unmanned Aerial Vehicle (UAV) Small Targets Based on Group Convolution" *Applied Sciences* 12, no. 11: 5402.
https://doi.org/10.3390/app12115402