A Comprehensive Study of Deep Learning and Machine Learning Methods for Land Application and Cover Segmentation

Research in the Delhi NCR area from 2000 to 2024 is compared in this study. Firstly, a flash flood occurred due to the heavy rain in July 2013. Grasslands are among the vegetative types in L ULC, which cannot be univocally classified as either natural in-land or artificial urban. But some areas of grassland, when it came time to classify L ULC, three different machine learning techniques were considered: random forest, support vector machine, and feedforward neural networks. Optimization The model best for random forest performed better than those two others. The study used statistical analysis tools, the “sustainable development prism” and “patch-corridor matrix” models, a landscape index, and time series data to explore how BGS changed in new urban areas. The research made use of Landsat datasets and optimized machine learning algorithms to classify L ULC, extract built-up areas, analyse landscape fragmentation and investigate built-up expansion in Delhi NCR by means of frequency analysis. In the EFPCNNM ensemble deep learning model, convolutional neural networks, feature pyramids, and ensemble learning are combined to increase the accuracy of remote sensing picture classification. The EFPCNNM process includes everything necessary for image analysis: pre-processing, feature extraction, feature pyramid creation, and ensemble learning. Exciting new possibilities for environmental monitoring, urban planning and property mapping emerge from this study. Sustainable urban development and resource management require effective monitoring and interpretation of the changes in L ULC, as well as employing remote sensing data alongside machine learning and deep learning technology.