Deep Learning for Detection of Visible Land Boundaries from UAV Imagery

Abstract

Current efforts aim to accelerate cadastral mapping through innovative and automated approaches and can be used to both create and update cadastral maps. This research aims to automate the detection of visible land boundaries from unmanned aerial vehicle (UAV) imagery using deep learning. In addition, we wanted to evaluate the advantages and disadvantages of programming-based deep learning compared to commercial software-based deep learning. For the first case, we used the convolutional neural network U-Net, implemented in Keras, written in Python using the TensorFlow library. For commercial software-based deep learning, we used ENVINet5. UAV imageries from different areas were used to train the U-Net model, which was performed in Google Collaboratory and tested in the study area in Odranci, Slovenia. The results were compared with the results of ENVINet5 using the same datasets. The results showed that both models achieved an overall accuracy of over 95 %. The high accuracy is due to the problem of unbalanced classes, which is usually present in boundary detection tasks. U-Net provided a recall of 0.35 and a precision of 0.68 when the threshold was set to 0.5. A threshold can be viewed as a tool for filtering predicted boundary maps and balancing recall and precision. For equitable comparison with ENVINet5, the threshold was increased. U-Net provided more balanced results, a recall of 0.65 and a precision of 0.41, compared to ENVINet5 recall of 0.84 and a precision of 0.35. Programming-based deep learning provides a more flexible yet complex approach to boundary mapping than software-based, which is rigid and does not require programming. The predicted visible land boundaries can be used both to speed up the creation of cadastral maps and to automate the revision of existing cadastral maps and define areas where updates are needed. The predicted boundaries cannot be considered final at this stage but can be used as preliminary cadastral boundaries.

Revising Cadastral Data on Land Boundaries Using Deep Learning in Image-Based Mapping

“…In general, deep learning is a relatively new research area in the geospatial domain and offers great potential for feature recognition from remote sensing imagery [30]. The upscaling deep learning solutions, including CNNs, for visible land boundary detection is becoming increasingly important, especially for UAV-based cadastral mapping [27,29]. Deep learning requires processing a large amount of training data and powerful computations.…”

“…Typically, CNNs are trained from scratch or by transfer learning. Both approaches require the preparation of custom training data, including images and labels, which usually takes some time and has already been highlighted in [27][28][29]. However, the amount of training data depends on the type of CNN architecture used.…”

“…Modern methods for automatic boundary detection in cadastral mapping also include deep learning [27][28][29][30]. Recent studies indicate that deep learning such as convolutional neural networks (CNNs) ensure a higher accuracy in delineating visible land boundaries than some of the state-of-the-art machine learning or object-based techniques [27,28].…”

“…The model was trained from scratch using hyperspectral UAV imagery. However, providing thousands of UAV training data can be considered a limitation, especially when a model is trained from scratch, as highlighted in [29]. This can be overcome with a CNN architecture that requires less training data, for instance, U-Net, followed by data augmentation prior to processing [29,33].…”

“…However, providing thousands of UAV training data can be considered a limitation, especially when a model is trained from scratch, as highlighted in [29]. This can be overcome with a CNN architecture that requires less training data, for instance, U-Net, followed by data augmentation prior to processing [29,33]. The challenge here remains to confirm what should be a sufficient variety of UAV training data to learn a robust network model.…”