Streamlining Testing for CNN model

Is there a way to have a python script grab all the images in a directory, and test the model (already trained model) with the images? Instead of just testing each image one by one, is there a faster way? Sort of like creating a test harness for the model, if that makes sense.