I am looking to test my own image on an object detection task with a trained model as this is described in Part 2 Lesson 9
, (Pascal VOC - object detection).
So far I haven’t seen any code to use a test set or even a single image to detect objects if any. Some specs about that image(s):
- should not be part of the training, i.e. it should not be part of the
md.trn_ds
neither themd.val_ds
. - there should be no ground-truth boundary box for it.
Has anyone tried it?