In order to ease the creation of multimedia annotations with our ontology, we have developed a Java API which provides an MPEG-7 class interface for the construction of meta-data at runtime. Annotations which are generated in memory can be exported to Java based RDF triple stores such as Sesame. For that purpose, the API translates the objects of the MPEG-7 classes into instances of the COMM concepts. The API also facilitates the implementation of multimedia retrieval tools as it is capable of loading RDF annotation graphs (e.g. the complete annotation of an image including the annotation of arbitrary regions) from a store and converting them back to the MPEG-7 class interface.
Download the API
COMM API incl. Javadoc
COMMAPI_SESAME incl. Javadoc
Usage Example
Using the API, the face recognition web service will automatically create the annotation which is depicted in Figure 2-F by executing the following code:
Image img0 = new Image();
StillRegion isr0 = new StillRegion();
img0.setImage(isr0);
StillRegionSpatialDecomposition srsd1 = new StillRegionSpatialDecomposition();
isr0.addSpatialDecomposition(srsd1);
srsd1.setDescription(new SegmentationAlgorithm());
StillRegion srr1 = new StillRegion();
srsd1.addStillRegion(srr1);
SpatialMask smr1 = new SpatialMask();
srr1.setSpatialMask(smr1);
RegionLocatorDescriptor rld1 = new RegionLocatorDescriptor();
smr1.addSubRegion(rld1);
rld1.setBox(new Rectangle(300, 230, 50, 30));
Semantic s1 = new Semantic();
s1.addLabel("http://en.wikipedia.org/wiki/Winston_Churchill");
s1.setDescription(new SVMClassifier());
srr1.addSemantic(s1);