Land Cover Classification (Landsat 8)

Land cover describes the surface of the earth. Land cover maps are useful in urban planning, resource management, change detection, agriculture, and a variety of other applications in which information related to earth surface is required. Land cover classification is a complex exercise and is hard to capture using traditional means. Deep learning models are highly capable of learning these complex semantics and can produce superior results. Using the model Follow the guide to use the model. Before using this model, ensure that the supported deep learning libraries are installed. For more details, check Deep Learning Libraries Installer for ArcGIS. Fine-tuning the model This model can be fine-tuned using the Train Deep Learning Model tool. Follow the guide to fine-tune this model. Input Raster, mosaic dataset, or image service. (Preferred cell size is 30 meters.) Output Classified raster with the same classes as in the National Land Cover Database (NLCD) 2016. Note: The classified raster contains 20 classes based on a modified Anderson Level II classification system as used by the National Land Cover Database. Applicable geographies This model is expected to work well in the United States. Model architecture This model uses the UNet model architecture implemented in ArcGIS API for Python. Accuracy metrics This model has an overall accuracy of 77 percent. The table below summarizes the precision, recall and F1-score of the model on the validation dataset. Class Collection 2 Level 2 Imagery Collection 1 Level 1 Imagery Precision Recall F1 Score Precision Recall F1 Score Open Water 0.96 0.97 0.96 0.95 0.97 0.96 Perennial Snow/Ice 0.86 0.69 0.77 0.49 0.94 0.64 Developed, Open Space 0.51 0.38 0.44 0.43 0.38 0.4 Developed, Low Intensity 0.52 0.46 0.49 0.47 0.48 0.47 Developed, Medium Intensity 0.54 0.5 0.52 0.49 0.54 0.51 Developed, High Intensity 0.67 0.54 0.6 0.55 0.68 0.61 Barren Land 0.76 0.59 0.66 0.6 0.77 0.68 Deciduous Forest 0.74 0.81 0.78 0.78 0.76 0.77 Evergreen Forest 0.77 0.82 0.79 0.8 0.82 0.81 Mixed Forest 0.56 0.47 0.51 0.5 0.53 0.51 Shrub/Scrub 0.82 0.82 0.82 0.84 0.81 0.83 Herbaceous 0.78 0.74 0.76 0.79 0.77 0.78 Hay/Pasture 0.7 0.74 0.72 0.67 0.75 0.71 Cultivated Crops 0.87 0.91 0.89 0.91 0.9 0.9 Woody Wetlands 0.7 0.68 0.69 0.67 0.68 0.68 Emergent Herbaceous Wetlands 0.72 0.54 0.62 0.54 0.61 0.57 Training data This model has been trained on the National Land Cover Database (NLCD) 2016 with the same Landsat 8 scenes that were used to produce the database. Scene IDs for the imagery were available in the metadata of the dataset. Sample results Here are a few results from the model.