Presented at the 4th International Symposium on Innovative Approaches in Engineering and Natural Sciences (ISAS WINTER-2019 (ENS)), Samsun, Turkey, Nov 22, 2019

SETSCI Conference Proceedings, 2019, 9, Page (s): 416-419 , https://doi.org/10.36287/setsci.4.6.107

Published Date: 22 December 2019    | 1096     18

 In today's applications, a significant increase in the use of deep learning algorithms is noticeable. The convolution neural network (CNN) of deep learning has been used frequently recently, especially for the successful discrimination of people and vehicles from other objects. Especially with the ImageNet Large Scale Visual Recognition Challenge (ILSVRC) in 2012, the use of CNNs has become widespread. With the development of technology and traditional image processing techniques, the proses of image processing has been considerably reduced, furthermore, the success rate has increased dramatically. Object detection can be difficult due to the low resolution of objects in aerial images. In this study, a system which automatically recognizes human and different types of vehicles (cars, bicycles, motorcycles) from aerial images taken with drone has been developed. In the system, Residual Networks (ResNet) model, which is the first in the ImageNet competition of the CNN been one of the deep learning techniques, is used. Google Colaboratory with Nvidia Tesla K80 GPU support is used for successful and fast training and testing of the system. In the developed system, results are explained according to different threshold values ​​of the objects detected from the images applied to the input.

Keywords - Deep Learning, CNN, Image Processing, ResNet, Drone

[1] Han, S., Shen, W., & Liu, Z. (2016). Deep Drone: object detection and tracking for smart drones on embedded system.
[2] Li, W., Li, H., Wu, Q., Chen, X., & Ngan, K. N. (2019). Simultaneously Detecting and Counting Dense Vehicles from Drone Images. IEEE Transactions on Industrial Electronics.
[3] Redmon, J., Divvala, S., Girshick, R., & Farhadi, A. (2016). You only look once: Unified, real-time object detection. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 779-788).
[4] Julia, D. L. f. (2016). "devblogs.nvidia.com." from https://devblogs.nvidia.com/parallelforall/mocha-jl-deeplearning-julia/
[5] Krizhevsky, A., Sutskever, I., & Hinton, G. E. (2012). Imagenet classification with deep convolutional neural networks. In Advances in neural information processing systems, 1097-1105.
[6] R. Girshick, J. Donahue, T. Darrell and J. Malik. (2014). Rich feature hierarchies for accurate object detection and semantic segmentation. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR).
[7] Karpathy, G. Toderici, S. Shetty, T. Leung, R. Sukthankar, and L. Fei-Fei. (2014). Large-scale video classification with convolutional neural networks. In Computer Vision and Pattern Recognition (CVPR), 2014 IEEE Conference on, 1725–1732. IEEE.
[8] N. Wang and D.-Y. Yeung. (2013). Learning a deep compact image representation for visual tracking. In Advances in Neural Information Processing Systems, 809–817.
[9] Dong, C., Loy, C. C., He, K., and Tang, X. (2015). Image super-resolution using deep convolutional networks. IEEE transactions on pattern analysis and machine intelligence, 38(2), 295-307.
[10] Girshick, R., Donahue, J., Darrell, T., & Malik, J. (2014). Rich feature hierarchies for accurate object detection and semantic segmentation. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 580-587).
[11] Bayram, F. (2020). Derin Öğrenme Tabanlı Otomatik Plaka Tanıma. Politeknik Dergisi.
[12] Özkan, İ. N. İ. K., & Ülker, E. (2017). Derin Öğrenme ve Görüntü Analizinde Kullanılan Derin Öğrenme Modelleri. Gaziosmanpaşa Bilimsel Araştırma Dergisi, 6(3), 85-104.
[13] Cengil, E., & Çınar, A. (2016). A New Approach for Image Classification: Convolutional Neural Network. European Journal of Technique, 6(2), 96-103.
[14] He, K., Zhang, X., Ren, S., & Sun, J. (2015). Delving deep into rectifiers: Surpassing human-level performance on imagenet classification. In Proceedings of the IEEE international conference on computer vision (pp. 1026-1034).
[15] Mikami, H., Suganuma, H., Tanaka, Y., & Kageyama, Y. (2018). Imagenet/resnet-50 training in 224 seconds. arXiv preprint arXiv:1811.05233.
[16] Goyal, P., Dollár, P., Girshick, R., Noordhuis, P., Wesolowski, L., Kyrola, A., ... & He, K. (2017). Accurate, large minibatch sgd: Training imagenet in 1 hour. arXiv preprint arXiv:1706.02677.