3
I wonder if you have knowledge and if there is a framework for recognizing objects, people, anyway, things, ready to be used on Android.
It’s similar to the idea of the app Google Goggles and the recognition of people on Facebook to mark them in a given photo.
If there is availability I suggest that not only recommend this tool, but also further detail the answer with documentation and example of use, so that the content is richer.
1
To recognize objects and a little complicated but not long ago Google made available in one of the Google Play services updates the Mobile Vision API where some of the tools contain face detection mechanisms, if they are smiling, the level of smile and unique characteristics of that person.
See the official documentation here
Find Objects in photos and video, using real-time on-device Vision technology. Source: Google. So I believe I can serve here too, I will research on, thanks.
Browser other questions tagged java android image image-processing
You are not signed in. Login or sign up in order to post.
There’s some specific object you want to recognize?
– gato
I need you to recognize animals, but taking into account color, stains, etc. The user will take the photo and will be compared to those already registered in the database.
– Giancarlo Abel Giulian
The subject is very broad, but you can use Simplecv is a framework that implements Opencv and other libraries that support Visual Computing, so you can create a server and do the recognition of the animas on the server and send the result to your App, I believe that if u save the images on the smarthphone locally and leave to do all the local processing will get very heavy.
– gato
Thank you very much, I’ll look him up.
– Giancarlo Abel Giulian
Object recognition is relatively smooth because they have a well-defined configuration. You can use Opencv, as my colleague @Denercarvalho has already mentioned. Here’s a nice tutorial for an important algorithm in the area: http://coding-robin.de/2013/07/22/train-your-own-opencv-haar-classifier.html
– Luiz Vieira
Face recognition is also feasible, especially in controlled environments (where you ask the user to position the face properly). In the Opencv documentation there are also examples: http://docs.opencv.org/2.4/modules/contrib/doc/facerec/facerec_tutorial.html
– Luiz Vieira
Now, recognizing animals is gonna be really hard. Firstly because there is no way you can control how the exposure will be (it is no use asking a dog to stare at the camera and not turn his head, for example). In addition, the variation of shape and color is huge.
– Luiz Vieira
I hope the tips are useful. Anyway, I also think the question is very broad (although it is a very interesting subject).
– Luiz Vieira
This msm Luiz, not how to control exposure, he would need a powerful algorithm that would be able to detect the animals in various angles, but even so arose more obstacles, in the case of birds, the algorithm would identify birds in a generic way and would not know which species is, for example, to differentiate a pigeon from a swallow. I believe that with the implementation of some neural network this would be possible, but there may be other ways to achieve the desired results that I do not know.
– gato
I really appreciate the comments, they helped me a lot to have an idea of how to do, I would imagine that with animals it would not be easy even more than I do not know about neural nets at the moment.
– Giancarlo Abel Giulian
@Deep Learning denercarvalho has been used even for this type of more subtle detection (say). : ) A colleague gave this answer with some tips on how to search (and there is a link from a Google tutorial that may be useful): http://answall.com/a/104778/73
– Luiz Vieira