Image capture and analysis on Android

Google Android™ is an operating system designed for mobile devices (phones, tablets, etc.). In 2005, Google bought a small startup called Android Inc. that was developing applications for cell phones. The aim of this takeover was to create a new mobile platform competing with the main players at the time: Symbian, Windows mobile, Palm OS, etc.

In 2007, the OHA (Open Handset Alliance) consortium, built around this system, was created. It brought together telephone operators, semiconductor manufacturers, handset manufacturers, software publishers and distributors. Android™ was then officially announced as the new platform for mobile devices.

In 2008, the first version of the SDK (Software Development Kit), was released. Since then, Androïd™'s market share has continued to grow, competing with Apple's (iPhone OS). Its fields of application are expanding by the day, and now even include home automation or TV sets.

Android™ is based on a Linux kernel. Its code is open and available as Open source. Its programming language is Java.

The success of Android™ has a lot to do with improving the user's interactivity with his or her device. This has been greatly enhanced by the arrival of new types of sensors.

The photo sensor is one of the many sensors present on any Android™ smartphone. It is an electronic component responsible for collecting light from the lens and converting it into electrical information. The image processing processor then converts this analog signal into digital information.

The aim of this article is to explain how to retrieve the characteristics of this sensor, and then show how to exploit them. In particular, we'll learn how to display the image coming from the sensor, how to capture it, and how to analyze it, thanks to various pre-trained models. Then we'll learn how to create our own using machine learning.