There are about three bits to that particular: 1. A work to create a database and this info exactly about this new pages you’ve liked and disliked. dos. A features to practice a product on the database. 3. A work to make use of this new educated model to help you instantly such as and you will dislike brand new users.
How it works
The last coating out-of an effective CNN instructed to own face class can be be taken given that an element place and that relates to a person’s face. It just thus happens that element lay is comparable to facial attractiveness.
tindetheus why don’t we your generate a database in accordance with the users one you adore and you may hate. You can then show a definition design into the database. The new design studies very first uses a MTCNN in order to position and you can field the new confronts on your own database. Then an effective facenet design is run using the brand new face to recoup the fresh embeddings (last coating of CNN). A logistic regression design will be match towards the embeddings. The brand new logistic regression model is actually stored, and this procedure try constant for the automation so you can automatically such and you may dislike profiles according to your own historical liking.