Have you ever wondered how Snapchat places that glass exactly on the bridge of your nose or that doggy nose at the tip of your nose? Human Vision makes it look like a very simple thing.
All you have to do is identify your nose and place the doggy nose there. But then how simple or complex is it for a machine to identify exactly the tip of 16 million unique users and place it exactly at the tip? It has to be taught to recognize patterns in the same way as human brains do. This is what Machine Learning does. It is a branch of AI that analyses data, builds a model and then predicts accordingly. Snapchat has been trained by a model that analyzed a number of faces, studied the patterns, and finally identified a specific facial feature from an image.
Google recently launched TensorFlow Lite which allows mobile developers to use AI on the mobile. This is a light weight solution that comes with the following models that has been trained and optimized for mobile.
Mobile Net: Vision Model that has been trained to detect more than 1000 different object classes.
Inception V3: An Image Recognition Model for detecting dominant objects present in an image.
Smart Reply: A Conversational Model that provides one touch replies for an incoming text message by suggesting contextually relevant messages. This has been successfully used in Android Wear 2.0 for providing Smart Messaging.
One can easily retrain these models on their own image datasets through transfer learning.
So how is it going to help the Landscape of App Development ?
With the launch of TensorFlow Lite, it’s going to be very easy for the developers to create apps with the following features:
- Speech Recognition and Translation
- Visual Search
- Object Detection
- Facial Recognition
- Landmark Detection
- Photo Editing apps
With the introduction of TensorFlow Lite, it becomes easier to adopt the capabilities of Machine Learning. At Ideas2IT, we have always been early adopters of great technologies. We have invested in Machine Learning across industries like banking and financial services, to develop mechanisms for lead scoring and more. If you’re looking to develop a mobile application with ML features, look no further.