Description: <DIV STYLE="text-align:Left;"><DIV><P><SPAN>THIS FEATURE CLASS CONTAINS PRELIMINARY STRUCTURE (BUILDING) POLYGONS FOR THE state of Maryland. The structures (buildings) are represented by polygon geometries with an attribute table described elsewhere in the metadata. WorldView-02 and WorldView-03 high resolution multispectral imagery was obtained for Maryland. The raw imagery (1B product) for Maryland was then pan-sharpened and orthorectified. Complete coverage of Maryland with minimal cloud cover includes imagery acquired between 2/10/2011 and 11/4/2019 and spatial resolutions between 0.46-0.85 meters. Once the imagery pre-processing was complete, training data was acquired and multiple convolutional neural network (CNN) models were generated. High Performance Computing (HPC) environments at ORNL were then used to apply the CNN to the imagery resulting in this preliminary structure (building) feature class. Minimal review of this dataset has been completed. Verification and validation of this dataset was conducted by an algorithmic model. See ‘Automated’ domain value for VAL_METHOD for more details. Only structure polygon results whose area is greater than 450 square feet are included in this feature class. The attribute schema associated with this feature class has been revised from previous data submissions. Population of some attributes is incomplete and identified in the Entity and Attribute Information section of this metadata. The visual representation of the features in this feature class have been improved.</SPAN></P></DIV></DIV>
Service Item Id: ccc0709a59ed4d1383bce10f8e8ddbdc
Copyright Text: Oak Ridge National Laboratory (ORNL)