0
$\begingroup$

I am collecting data to train an object detection model using and was wondering if 5 labels in the same image and 5 images with 1 label each provided the same quality of input training data. Example: an image with 5 labeled apples vs. 5 images with 1 apple each.

$\endgroup$
1
  • $\begingroup$ If I may ask, why '5 images with 1 label each'? Technically speaking, having setup the dataset like this may lead to intentional duplicates in your data (somewhat, since your labels are changing, you may have less problem, but it is exactly same image), and in general duplicates introduces bias to your data! You may want to try it out to compare on a set of unseen dataset. $\endgroup$ – TwinPenguins Apr 28 '20 at 5:49
0
$\begingroup$

No. These are two separate problems. Multi-label classification and multi-class classification. In general, when we talk about classification we mean multi-class classification i.e. there are a certain number of categories and the input training samples fall into only one of these which is your case of 5 images with 1 label. In the case of 1 image with 5 labels, it is a multi-label classification, which is a generalization of multi-class classification. Both of these are different tasks and the use case depends on the problem that you are trying to solve!

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.