tangyuhao
commented
7 years ago you should generate the training data using exactly same method described in the GOTURN papaer. the first searching region should be defined manually, there is no method to track an object in an video if you don't tell it what to track. preprocessing the dataset is really big pain and complicated, if you want to do that please refer to the GOTURN paper.
lidongxuan
I see the test data you given, how could I generate the training data ? For example, I have the video and corresponding bounding box location, I know the target object, how could I set the searching region ? Random setting the bounding box ??? It will be cool if you could shown how to pre-processing the dataset ! Thanks !