1

I am training an object detector using mxnet/resnet50 After the last training run the mAP was 78%, and the loss was 0.37 When I run the detector on my test set (independent of train/val data) I am getting false positives result - with some rather high 30-60% confidence levels. I think I need to add some train/val images that do not have ANY of the objects i'm training the detector for.

I'm planning on adding about 20% more images that have a label of -1 -- which I read somewhere is how you designate an image having no label in mxnet.

Does this seem reasonable? is -1 the right way to designate it? any downside? Thanks, john

jlwebuser
  • 49
  • 7
  • In theory any unused category would work, including -1. Are you following a specific example I can comment on? What data set are you currently using, and how many categories does it have with objects? Adding training data without objects is the way to go, but finding the sweet spot of how much to add relative to the data with classes is effectively a hyperparameter. – Vishaal Jan 16 '19 at 01:00
  • Hi - following up - can you include more specific info so we can answer the question? Thanks! – Vishaal Mar 01 '19 at 22:53

1 Answers1

0

One method for an unbalanced object detection task is to have a classifier before the object detection stage, which determines if the image contains an object or not. You can weight the loss for each class in this classifier relative to its inverse frequency (i.e. higher weight for classes that appear less frequently). You should test on data with a similar class balance as the real world. You might find this post useful.

Thom Lane
  • 993
  • 9
  • 9