There are a number of commercial APIs such as Clarifai and Google Cloud Vision that provide image tagging, however they don't provide any location or size (bounding box) for the detected objects, they simply just provide tags for the image as a whole.

Is there any easy to use open source solution (but preferably a commercial API) that can detect objects (and doesn't need to be trained - can detect thousands of every day objects out of the box) and return their bounding box as well, similar to this image:

enter image description here



  • I don't have time to write a full answer but I recommend caffe – Hack-R Dec 24 '16 at 15:31
  • I've looked at caffe and it does look interesting, however as mentioned above, I am looking for a solution that can show where the tags appear in the image - and caffe doesn't seem to have that localization feature...unless I've missed something. – abagshaw Dec 24 '16 at 19:54
  • Oh sorry yes I missed that. You could do it but it would take custom coding. I think I did something close to that a year ago but it was using tensorflow and not easy at all. I suggest you ask on Kaggle and / or the caffe listserv. – Hack-R Dec 24 '16 at 21:07

Your Answer

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

Browse other questions tagged or ask your own question.