1 | initial version |
Then you need both the original data and your new data. Add your new detections as hard positives at the beginning of your positive training set to enforce them being used and to ensure that they add more knowledge to the model.