BOW number of words in vocabulary for unique images database?

edit

@Guanta, I don't think I misunderstood BOWKMeansTrainer. It clusters train descriptors into words for the vocabulary, given how many clusters you want. Also, repeating again, I do not want to classify images. I want to leverage the ease of use and fast indexing from the inverted index approach in BoW to find out which clusters the query images falls most into, then find those clusters' associated database images, and then find out which image has the highest frequency of occurrence from the lookup. Maybe this would make it more clear to you: http://www.cs.utexas.edu/~grauman/courses/fall2009/slides/lecture16_bow.pdf goto slide 25 in it, labelled inverted index. Edit: Thanks for answering though, @Guanta.

Also, maybe I missed something: my main motive is to have BoW inverted index replace my current FLANN based image index. Because the flann matcher I use to match query image with many images currently is not savable or updatable (one cannot simply add another image's descriptors and use it without retraining to use it for matching, and deleting a particular set of descriptors is not possible). And I want to save it to disk, not generate it from saved descriptor data everytime my application starts.

Okay, it got clearer now what you like to do. So, the BoW inverted index is typically drawn from the same clusters which you used to create the BoW-descriptors, since typically you use the index file for weighting the BoW descriptors. Of course you could also cluster the local descriptors differently and use thus a different number of clusters, but why? I would go with the simple way.

For updating: I wouldn't retrain the image database and just update the inverted index file, if your training images were diverse enough then you are good. If you would retrain it, you'd also need to recreate the inverted index file which is I guess pretty time consuming.

I don't understand this part: Of course you could also cluster the local descriptors differently and use thus a different number of clusters, but why? What local descriptors? The one I take from the image by SURF/ORB ? And differently how? Actually, I just don't understand any of it. Sorry.

And this: I would go with the simple way. Are you referring to the FLANN based image index which I mentioned above or else what?

And then: For updating: I wouldn't retrain the image database and just update the inverted index file You mean I just add the SURF descriptors of a new image added to my server to my BoW trainer object and then somehow without retraining it to cluster the new set of descriptors into the existing clusters, I should update my inverted index? Seems crazy.

For the last part: How would I even update the vocabulary to include the new set of descriptors from the newly added image without retraining the BOWKMeansTrainer object to get a new vocab? And then only can I get BOW descriptor of the new image with BOWImgDescriptorExtractor and use the result to update the inverted index. Please correct me if I'm missing something from the flow.

"what local descriptors" -> yes SURF/ORB/Whatever you have chosen here, these are local descriptors, the BoW descriptor for an image is called the global descriptor. You need to cluster the local descriptors to create the vocabulary (aka dictionary), this vocabulary is then used to a) create the BoW descriptor for each image (each image, it doesn't need to be in the training set) b) it is typically also used to create the inverted index file to weight then the BoW descriptors. However, if you don't want to use it for weighting then you could in theory also train it differently, also with different numbers of clusters. But as I wrote, I wouldn't do it and just use the same clusters as before, i.e. only one clustering is involved.

"you mean I just add the SURF descriptors" -> no retraining (only occasionally if flann decides to do so). You compute your BoW descriptor as before, i.e. create histogram of all SURF descriptors to its nearest clusters. With updating I meant only updating the inverted index file: for each cluster where you raise the bin count you also save the image index (this is how you create the inverted index file), so you just need to update it with the new infos. This is possible, e.g. in flann w. addPoints()-method, however you are right in the sense that the flann index may be needed to be retrained (depends on the parameter you used with the addPoints()-method).

For your last comment: I think you still mix some parts up: you don't need to retrain the BoWKMeansTrainer() (=vocabulary), this is needed only once (if your new images are not too much different from the images used when training it)!

I think you still mix some parts up: you don't need to retrain the BoWKMeansTrainer() (=vocabulary), this is needed only once (if your new images are not too much different from the images used when training it)!

That's the thing dude! All my images will always be different so much so that atleast I as a person even wouldn't saw that any two of them belong in a same class or can be classified as "similar".

So, for every image added to my server, I will need to: 1) Update the vocab with the descriptors of those new images. 2) Get the nearest clusters for those new images and then add them to inverted index.

Continuing..

Now, without 1, 2 is not possible, or, say, how does it make sense to match a query image with a matcher that isn't even trained with the target image's descriptors? (because we didn't add descriptors of the new image to bowtrainer and then call cluster on it)