Object Based Terrain Classification Using Deep Learning Techniques
Suneetha Manne, Anguluri Venkata Siva Pavani Sujitha, Mohammad Shameera, Movva Ram Kalyan · 2024
In the realm of terrain classification, the motivation behind this project lies in addressing the persistent challenges associated with accurate mapping of diverse landscapes. Motivated by the need for advanced techniques in terrain classification, this project addresses the challenge of developing a robust deep learning model, specifically a Convolutional Neural Network (CNN). The goal is to enable accurate recognition and categorization of diverse terrain types through the analysis of visual information extracted from images. Existing methodologies often fall short in achieving precise terrain classification, leading to suboptimal decision-making in various domains. The proposed model aims to overcome these limitations by employing state-of-the-art architectures like the 19-layer Visual Geometry Group (VGG19) and Residual Network with 50 layers (ResNet50). The significance of this work lies in the improved accuracy achieved by the models, with VGG19 and ResNet50 demonstrating impressive accuracies of 97.67% and 97.59%, respectively. These outcomes underscore the effectiveness of deep learning in addressing the complexities associated with terrain classification. The results not only highlight the exceptional capabilities of the proposed models but also emphasize their potential applications in diverse fields. The potential uses span from monitoring the environment and responding to disasters to urban planning, where precise terrain classification plays a crucial role. Fundamentally, this initiative presents a groundbreaking answer to the current hurdles in terrain classification, supplying a potent tool with extensive relevance and significant societal influence.