ToyRecognizer with OpenCV

Are you parents, overwhelmed by toys whose name is very difficult to remember? Your little kid likes to invent names for their favourite little friends? Your kids like to draw and name the characters in the drawing? You have a hard time keeping up with the imagination of your offspring??? Your babysitter does not know how to name the 8 dwarfs, 6 gnomes and 4 wizards of your kids collection? There are 20 princesses in 4 different castles that need to saved by 20 princes and related horses???

Then ToyRecognizer is the tool for you!

ToyRecognizer is an OpenCV application that can recognize toys or drawings by means

of computer vision algorithms. The application will match a photo taken by the user with the

5 closest descriptions available in the internal database of the application, in which the user has previously placed

the toys and their description.

When the user selects one of the descriptions, an image will be produced, visualizing the result description. The user can then navigate the obtained results.

The application allows to store images of toys taken by the user, indifferently from whether the model is a drawing or an action figure. The application asks the user to provide a description associated to the new inserted model in the ToyRecognizer database.

The application depends slightly on the illumination and on the background used for the photos, as it may take keypoints for the matching anywhere in the figure. Pay attention to make sure that the background is not more complex than the main subject in the photo, or probably the results obtained will not be very accurate. Our advice is to position the toy on a monochromatic background (possibly a white/grey/black sheet).

The application does not store user data, it is for free, but it runs advertisements, so it requires the permission to use the internet connection.