Mobile application that can find a celebrity you have a strong resemblance to.
Task:
Create a nowadays modern celebrity face matching app.
Technologies and Tools:
Server: Python, OpenCV.
Client: Swift, Objective-C, C++, TensorFlow, OpenCV, DNN, Google Firebase Vision.
Challenge
Our task was to create a mobile application that compares a user’s photo with celebrity faces and suggests the most similar options. The inspiration was an application called Gradient – mainly we focused on the work and functionality of this particular product.
Solution
Photos’ graph learning aspect with using of TensorFlow on the server side of the app allows finding the needed similarity.
Users select a photo that they want to interact with. Then, the system searches for a maximum match and gets a name of similar сelebrity.
In the end we get 2 photos: user’s photo and photo of celebrity. For a complete picture and proof that the photos are really similar, we use Morphing effect.
Result
We got the application of a very exciting type that is now widely popular. Users can find a famous figure that they look a lot like by tapping a couple of buttons.
System detects a key components of user’s face that allows finding a really similar person. To get the better face recognition accuracy, the face should be clearly visible.