It seems that raster predict function can pass to predict.randomForest by setting the "type='response'". I am using a parallel version of randomForest package called the "randomForestSRC" and I am hoping to pass the raster predict function to predict.rfsrc to take the advantage of the parallelization and low memory usage. Does anyone know if raster predict function supports this? Or if there is a way to use raster predict function to do this? Or does anyone know if there are any other packages that can predict a fitted model to raster data and have the ability to handle memory issues (since I am using huge satellite image data)? .