DeepLearning4J Service vs OpenCV DL4J filter vs Yolo

Ahoy Kwatters !
I was wondering the differences and the plan going forward for these 3.

OpenCV DL4J Filter - Worky !

OpenCV Yolo Filter - Worky !

Deeplearning4J Service has no UI in swing or web - but I can see the filter spawns the service

In a filter it would seem Yolo is more useful since it provides segmentation / localization of images found ... also seems to be less cpu intensive.  I'm guessing its because DL4J is a general dnn service and the filter uses a model that was trained for image recognition, but Yolo was designed from its foundation to do recognition/segmentation.

I'm asking because I'm wondering if a web ui should be created fro DL4J - ah .. I see some tranfer learning stuffs in there and solr references - heh, harkens back to the fun times in Montreal :)

The current errors in Jenkin builds 
"RuntimeException: problem during retrieve of org.myrobotlab#myrobotlab-core: java.lang.RuntimeException: Multiple artifacts of the module org.bytedeco#javacpp;1.5.4 are retrieved to the same file! Update the retrieve pattern to fix this error."
come from an explicit dependency reference in Deeplearning4J service.

I experimented by taking the javacpp out of my pom, and it appeared to run fine - is the reference there for GPU hardware ?

(I'll continue to do testing/experimenting to shed more light)