Deep Learning-Based Comparison of Performance of Different Object Detection Models for Images Obtained at Various Elevations Using UAV

As the use of deep convolutional neural networks (DCNNs) for object detection has increased, methods for deep learning are developing quickly. In contrast to traditional handmade feature-based approaches, deep learning-based object recognition systems may learn both low-level and high-level picture properties. Deep learning-based picture features are more representative than manually constructed features. As a result, the focus of this paper is on deep convolutional neural network-based object detection techniques; however, conventional object detection algorithms will also be briefly discussed with an example. In this research, an automated method for detecting different trees from very high-resolution UAV imagery is addressed. The proposed method uses two models—YOLO v7 and Roboflow 2.0 and transfer learning using convolutional neural networks to identify “Eucalyptus” and “Neem” trees.