In ROI Annotation, currently raw ROI of ArrayList are passed in to feature extractor. This works only when we insist each feature extractor provide two entry methods, one for DataInput, one for arraylist. Need to revisit the possibility of consolidate them. What I can think of as the drawback is the additional DataInput object created during ROI annotation (it should be reused to avoid garbage object), but that should be justifiable for clean interface of FeatureExtractor.
In ROI Annotation, currently raw ROI of ArrayList are passed in to feature extractor. This works only when we insist each feature extractor provide two entry methods, one for DataInput, one for arraylist. Need to revisit the possibility of consolidate them. What I can think of as the drawback is the additional DataInput object created during ROI annotation (it should be reused to avoid garbage object), but that should be justifiable for clean interface of FeatureExtractor.