Derin Öğrenme ve Öğrenme Aktarımı Algoritmalarının Drone Algılama Performansı Üzerine Etkisi

Fatma Gülşah Tan

TR EN

Derin Öğrenme ve Öğrenme Aktarımı Algoritmalarının Drone Algılama Performansı Üzerine Etkisi

Öz

Drone teknolojilerinin hızla gelişmesiyle birlikte, özellikle güvenlik ve gizlilik alanlarında ciddi endişelere yol açan drone kullanımı, günümüzde büyük bir önem arz etmektedir. Derin öğrenme ve öğrenme aktarımı yapay zekâ teknikleri, drone tespiti konusunda umut vaat etmektedir. Ancak, bu tekniklerin başarıyla uygulanabilmesi için, karmaşık hava koşulları, değişken hızlar ve yüksek manevra kabiliyetine sahip droneların doğru şekilde saptanabilmesi için yeni ve verimli çözümler geliştirme ihtiyacı kaçınılmazdır. Bu çalışmada, drone nesnelerinin tespiti için EfficientNet modeli kullanarak eğitim modellerinin drone tespiti üzerindeki performansları ve karşılaşılan zorluklar karşılaştırılarak, gelecekteki potansiyel başarıları hakkında bir perspektif sunulmuştur. Elde edilen sonuçlara göre, öğrenme aktarımı yönteminde daha fazla katman dondurulduğunda, eğitim için gereken GPU belleği azalır ve GPU kullanımı düşer. Bu durum, daha büyük görüntü boyutlarıyla eğitilen modellerin daha hızlı eğitilebileceğini göstermiştir. Derin öğrenme yöntemi daha fazla veriye ve GPU kaynağına ihtiyaç duymaktadır, bu da eğitim süresini uzatmaktadır. Yapılan deneylerde derin öğrenme yöntemiyle eğitilen modelin en iyi başarı oranı %97.3, öğrenme aktarımı yöntemiyle eğitilen modelin en iyi başarı oranı ise %99.7 olarak belirlenmiştir. Bu, öğrenme aktarımı yönteminin az veriyle daha yüksek bir doğruluk oranı sağladığını göstermektedir. Ancak, derin öğrenme yöntemiyle elde edilen başarı oranı da oldukça tatmin edici bir sonuç olarak değerlendirilebilir.

Anahtar Kelimeler

EfficientNet , derin öğrenme , öğrenme aktarımı , nesne tespiti , drone tespiti

The Effect of Deep Learning and Transfer Learning Algorithms on Drone Detection Performance

Abstract

With the rapid advancement of drone technologies, the use of drones, particularly in the areas of security and privacy, has become a matter of great concern today. Deep learning and transfer learning artificial intelligence techniques hold promise in the field of drone detection. However, for these techniques to be successfully applied, there is an inevitable need to develop new and efficient solutions for accurately detecting drones in complex weather conditions, variable speeds, and high maneuverability. In this study, the performances of training models using the EfficientNet model for drone detection are compared, and the challenges encountered are discussed. A perspective on potential future successes is presented. According to the results obtained, when more layers are frozen in the transfer learning method, the GPU memory required for training decreases, and GPU usage drops. This indicates that models trained with larger image sizes can be trained faster. The deep learning method requires more data and GPU resources, leading to an extended training time. In the experiments, the model trained with the deep learning method achieved the highest success rate of 97.3%, while the model trained with the transfer learning method achieved the highest success rate of 99.7%. This demonstrates that the transfer learning method provides higher accuracy with less data. However, the success rate achieved with the deep learning method is also considered quite satisfactory.

Keywords

EfficientNet , DeepLearning , TransferLearning , object detection , Drone Detection

References

[1] G. Marques, D. Agarwal, and I. de la Torre Díez, “Automated medical diagnosis of COVID-19 through EfficientNet convolutional neural network,” Appl Soft Comput, vol. 96, p. 106691, Nov. 2020, doi: 10.1016/j.asoc.2020.106691.
[2] M. Tan and Q. V. Le, “EfficientNetV2: Smaller Models and Faster Training,” ArXiv, Apr. 2021. [Online]. Available: https://arxiv.org/abs/2104.00298. [Accessed: Sep. 25, 2023].
[3] A. K. Kakumani, “Deep Learning Architechture For Classification Of Breast Cancer Cells in Fluorescence Microscopy Images,” International Journal of Advanced Research in Computer Science, vol. 13, no. 06, pp. 44–44, Dec. 2022, doi: 10.26483/ijarcs.v13i6.6921.
[4] S. Al-Emadi, A. Al-Ali, and A. Al-Ali, “Audio-Based Drone Detection and Identification Using Deep Learning Techniques with Dataset Enhancement through Generative Adversarial Networks,” Sensors, vol. 21, no. 15, p. 4953, Jul. 2021, doi: 10.3390/s21154953.
[5] Z. TAN, M. KARAKÖSE, and E. ÖZET, “Drone Tracking with Drone using Deep Learning,” International Journal of Computer and Information Technology(2279-0764), vol. 11, no. 3, Aug. 2022, doi: 10.24203/ijcit.v11i3.238.
[6] F. Samadzadegan, F. Dadrass Javan, F. Ashtari Mahini, and M. Gholamshahi, “Detection and Recognition of Drones Based on a Deep Convolutional Neural Network Using Visible Imagery,” Aerospace, vol. 9, no. 1, p. 31, Jan. 2022, doi: 10.3390/aerospace9010031.
[7] Y. Wang, Y. Chen, J. Choi, and C.-C. J. Kuo, “Towards Visible and Thermal Drone Monitoring with Convolutional Neural Networks,” APSIPA Trans Signal Inf Process, vol. 8, no. 1, 2019, doi: 10.1017/ATSIP.2018.30.
[8] Z. Wu, S. Pan, F. Chen, G. Long, C. Zhang, and P. S. Yu, “A Comprehensive Survey on Graph Neural Networks,” IEEE Trans Neural Netw Learn Syst, vol. 32, no. 1, pp. 4–24, Jan. 2021, doi: 10.1109/TNNLS.2020.2978386.
[9] L. Aziz, Md. S. Bin Haji Salam, U. U. Sheikh, and S. Ayub, “Exploring Deep Learning-Based Architecture, Strategies, Applications and Current Trends in Generic Object Detection: A Comprehensive Review,” IEEE Access, vol. 8, pp. 170461–170495, 2020, doi: 10.1109/ACCESS.2020.3021508.
[10] T. Takahashi, K. Nozaki, T. Gonda, T. Mameno, M. Wada, and K. Ikebe, “Identification of dental implants using deep learning—pilot study,” Int J Implant Dent, vol. 6, no. 1, p. 53, Dec. 2020, doi: 10.1186/s40729-020-00250-6.

[11] Y. Fu, “Recent Deep Learning Approaches for Object Detection,” Highlights in Science, Engineering and Technology, vol. 31, pp. 64–70, Feb. 2023, doi: 10.54097/hset.v31i.4814.
[12] G.-S. Xia et al., “DOTA: A Large-Scale Dataset for Object Detection in Aerial Images,” in 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, IEEE, Jun. 2018, pp. 3974–3983. doi: 10.1109/CVPR.2018.00418.
[13] J. Zhang and D. Tao, “Empowering Things With Intelligence: A Survey of the Progress, Challenges, and Opportunities in Artificial Intelligence of Things,” IEEE Internet Things J, vol. 8, no. 10, pp. 7789–7817, May 2021, doi: 10.1109/JIOT.2020.3039359.
[14] N. Zhao, Z. Wu, R. W. H. Lau, and S. Lin, “What makes instance discrimination good for transfer learning?,” Jun. 2020. [Online]. Available: https://arxiv.org/abs/2006.06606. [Accessed: Dec. 12, 2023].
[15] S. Shende, “CNN Based Missing Object Detection,” Int J Res Appl Sci Eng Technol, vol. 11, no. 4, pp. 956–959, Apr. 2023, doi: 10.22214/ijraset.2023.50138.
[16] S. Agarwal et al., “Unleashing the power of disruptive and emerging technologies amid COVID-19: A detailed review,” May 2020. [Online]. Available: https://arxiv.org/abs/2005.11507. [Accessed: Sep. 25, 2023].
[17] I. Athanasiadis, P. Mousouliotis, and L. Petrou, “A Framework of Transfer Learning in Object Detection for Embedded Systems,” Nov. 2018. [Online]. Available: https://arxiv.org/abs/1811.04863. [Accessed: Sep. 13, 2023].
[18] F. Wang et al., “Recent Advances in Fatigue Detection Algorithm Based on EEG,” Intelligent Automation & Soft Computing, vol. 35, no. 3, pp. 3573–3586, 2023, doi: 10.32604/iasc.2023.029698.
[19] S. Shen et al., “K-LITE: Learning Transferable Visual Models with External Knowledge,” Apr. 2022. [Online]. Available: https://arxiv.org/abs/2204.09222. [Accessed: Sep. 13, 2023].
[20] I. T. Plata, E. B. Panganiban, D. B. Alado, A. C. Taracatac, B. B. Bartolome, and F. R. E. Labuanan, “A Recognition Method for Cassava Phytoplasma Disease (CPD) Real-Time Detection based on Transfer Learning Neural Networks,” International Journal of Advanced Computer Science and Applications, vol. 12, no. 12, 2021, doi: 10.14569/IJACSA.2021.0121234.
[21] M. M. Rahaman et al., “Identification of COVID-19 samples from chest X-Ray images using deep learning: A comparison of transfer learning approaches,” J Xray Sci Technol, vol. 28, no. 5, pp. 821–839, Sep. 2020, doi: 10.3233/XST-200715.
[22] M. E. H. Chowdhury et al., “Can AI Help in Screening Viral and COVID-19 Pneumonia?,” IEEE Access, vol. 8, pp. 132665–132676, 2020, doi: 10.1109/ACCESS.2020.3010287.
[23] C. Zhu, J. Liang, and F. Zhou, “(Retracted) Transfer learning-based YOLOv3 model for road dense object detection,” J Electron Imaging, vol. 32, no. 06, Jan. 2023, doi: 10.1117/1.JEI.32.6.062505.
[24] Saha, I., “drone-bird classification,”. [Online]. Available: https://www.kaggle.com/datasets/imbikramsaha/drone-bird-classification [Accessed: Dec. 25, 2023].

Details

Primary Language

Turkish

Subjects

Computer Software

Journal Section

Research Article

Authors

Fatma Gülşah Tan ^*
0000-0002-2748-0396
Türkiye

Publication Date

December 31, 2023

Submission Date

November 15, 2023

Acceptance Date

December 5, 2023

Published in Issue

Year 2023 Volume: 9 Number: 4

IZ

https://izlik.org/JA36DZ34AF

IEEE

[1]F. G. Tan, “Derin Öğrenme ve Öğrenme Aktarımı Algoritmalarının Drone Algılama Performansı Üzerine Etkisi”, GJES, vol. 9, no. 4, pp. 1–13, Dec. 2023, [Online]. Available: https://izlik.org/JA36DZ34AF

GJES is indexed and archived by:

33113 33114 33115 33116 33117

Gazi Journal of Engineering Sciences (GJES) publishes open access articles under a Creative Commons Attribution 4.0 International License (CC BY)