Landmark Detection Using Convolutional Neural Network: A Review

Geographic recognition is an important task in computer vision and plays an important role in many applications such as facial recognition, medical image analysis, and image management. Over the past few years, convolutional neural networks (CNN) have become an important method for spatial detection due to their excellent ability to capture spatial features. This review paper provides a comprehensive review of spatial sensing methods using CNNs, focusing on recent advances and challenges in this field. We then go into the basics of CNNs and their applicability to field surveys, highlighting their ability to learn hierarchical representations from objects. We discuss popular CNN models such as AlexNet, ResNet, and GoogleNet and explore their advantages and limitations in geographic research projects. The impact of various loss factors, including squared error and customized design loss, on field accuracy will be analyzed. To better understand progress in this field, we examine the most useful data and benchmarks used in field research. And throughout the paper we discuss recent changes and achievements such as the integration of tracking systems, the use of image-based methods, and the use of deep neural networks for 3D landmark detection. Finally, we describe current challenges and future directions for using CNNs for signature detection, regarding the need for closure, scalability, and robustness for translation. We propose a possible research method that involves searching various research sites and combining data sources.