International Science Community Association International Research Journal of Earth Sciences 14 2 2026 08 25 Review on Deep Learning Methods, and their Implications for Mapping Geomorphic Landforms 8 17 EN Issa Salifou 1 Multi-purpose Laboratoire Polyvalent, Diffa University, PO Box: 78 Diffa, Niger and Department of Water and Fisheries Resources, High Institute of Environment and Ecology, Diffa University, PO Box: 78 Diffa, Niger Anabi Toudjani Assane 2 Department of Water and Environment, Faculty of Agronomic and Ecological Sciences, Diffa University, PO Box: 78 Diffa, Niger Maman Maazou Ahmed 3 Faculty of Agronomic Sciences and Food Technologies, Boubacar Bâ University of Tillabéri (UBBTi), PO Box: 175 Tillabéri, Niger Adamou Zanguina 4 Department of Chemistry, Faculty of Science and Technology, Abdou Moumouni University, PO Box: 10662 Niamey, Niger Natatou Ibrahim 5 Department of Chemistry, Faculty of Science and Technology, Abdou Moumouni University, PO Box: 10662 Niamey, Niger Ali Muhammad 1 Department of Geology and Geography, West Virginia University, Morgantown, WV 26505, United States Asghar Hussain 2 Institute of Geology University of Azad Jammu and Kashmir, Muzaffarabad AJ&K, Pakistan > Review Paper 2026 1 26 2026 08 25 The importance of accurate mapping of geomorphic and archeological features has arisen across the globe after the great success of deep learning techniques. Deep learning models are novel approaches utilized for the mapping of geomorphic landforms, and archeological features to improve accuracy, reduce time, provide automation, and produce high-resolution maps. Convolutional Neural Networks (CNNs) have proven to be one of the most promising techniques applied effectively for the mapping of geomorphic landforms. Spatial and semantic parameters of the CNNs and U-Net are reviewed in this work, and their implications for the accurate mapping of geomorphic landforms. Additionally, very recently, a series of studies have established that Convolutional Neural Networks (CNN) are very effective in pixel-level classification and aggregated landforms. This review presents a few examples where CNN has shown accurate results. CNN has measured an ice-wedge polygon in Alaska with a 91% accuracy rate, and the CNN used as an object-based image analysis in the northern Netherlands to map low-relief features. Furthermore, this study presents a comprehensive deep insight review into CNN and its utilization for accurate mapping of complex and transitional landforms since pixel and object-based techniques do not perform well in classifying complex and transitional landforms. Copyright@ International Science Community Association