HN Explorer

Ask HN: How to handle no detection in Deep Learning?

2 points by dekhtiar ↗ HN
Hello dear friends,

I try to develop a Deep Learning Model for my PhD Work that detect a set of objects. And I heavily struggle on one point. How do you handle the fact that none of the "known objects" is detected ? I mean even if I input a black or white image, I still have an output such as "ObjectX: xx%". How to change it for a blank result or "no detection".

I obviously can't input millions of situations where there is no object on the image with the label "no image".

I hope my explanation will be clear.

Best regards and thansk for the help.

2 comments

[ 3.6 ms ] story [ 15.0 ms ] thread
visarga
You could train a discriminator on top of your classifier, and feed your discriminator with both kinds of images (with and without objects). The discriminator would benefit from the classifier and you wouldn't need too many examples.

A simpler way would be to output probabilities per label instead of just labels, then threshold those with probability larger than a certain value, say, 0.9.

Or you could try to compute the entropy of the prediction vector, and if the entropy is too high, consider it "undecided".

dekhtiar
Hello, Thanks a lot for your input.

A discriminator could be a great idea. Any idea on the kind of model efficient for such a task. I'm not sure a CNN is required for such a simple job.

Nowadays, I ouput label + probabilities, however detection threshold can sometimes be around 20% or 70%. I could of course design label-specific threshold, but it doesn't seem to be the correct way.

Or sometimes there is an object in front of me, but not any I know to classify. I would like to be able to detect that this object I don't know.

Concerning entropy, I would definitely give it a shot ! Very nice idea.

Any of you may know a system implementing such behaviour (e.g. I don't detect any object, I don't know this object) ?

Thanks a lot for your help.