I'm wondering if e-commerce companies where products are offered by users, such as EBay, are using Object Recognition to ensure that an uploaded image corresponds to an specific type of object (clothing, shoes, glasses, etc) either to classify automatically or more importantly to filter undesired images (such as non related or even illegal types).

If so, which algorithms and/or open platforms could be use for doing so? From what I've looked it seems that HOG+Exemplar SVM might be one of the most accurate methods developed so far (http://www.cs.cmu.edu/~efros/exemplarsvm-iccv11.pdf), even having couple of public repo's with Matlab implementations (https://github.com/quantombone/exemplarsvm), but I'm still wondering if this is being used in industry.

  • $\begingroup$ Know it is at least being researched (in case of Amazon) don't know state of implementation. $\endgroup$ – seanv507 Jan 1 '15 at 14:35

As far as I am aware, object recognition is not extensively used in industry yet. Google's image search, for instance, is based on exploiting the text on the web pages rather than the images themselves. I have seen several start-up companies that market prototypes based on object recognition, but these products are either prototypes or not widely used.

Regarding state-of-the-art techniques, you should take a look to recent literature on deep learning and deep neural networks applied to object recognition, along with more specific techniques like convolutional networks.

| improve this answer | |

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.