Enhancing Land Use Identification Through Fusion of Densenet121 DCNN and Thepade SBTC Features Using Machine Learning Algorithms and Ensembles

Aerial vehicles such as drones are used to capture high-resolution images are also called aerial images. Using these aerial images for the identification of land usage helps in remote sensing, geospatial analysis, and urban planning. Aerial images have proven to be an efficient and cost-effective means of monitoring changes in land usage and cover over time, providing valuable insights for various fields such as agriculture, forestry, and environmental management. The paper proposes the fusion of the auto-extracted feature of Densenet121 with the handcrafted colour feature of the aerial images using Thepade sorted BTC (TSBTC) for getting the performance of LUI. This paper uses eight ML algorithms and three top ensembles from ML algorithms. The experimental results demonstrate that ensembles outperform individual ML algorithms in achieving better LUI results. The results obtain from experimentation indicate that the fusion of features from TSBTC and Densenet121 yields better performance in LUI than using either TSBTC or Densenet121 individually.