Hi Lu,
The trainCascadeObjectDetector function creates an XML file containing the classifier parameters. In the code you've shown, the file will be named 'stopSignDetector.xml', so you should probably name it something else, like 'craterDetector.xml'. You can then create a vision.CascadeObjectDetector object using this XML file, as follows:
detector = vision.CascadeObjectDetector('craterDetector.xml');
