VID2META: Complementing Android Programming Screencasts with Code Elements and GUIs

The complexity of software projects and the rapid technological evolution make it such that developers often need additional help knowledge to tackle their daily tasks. For this purpose, they refer online resources, which are easy access contain a wealth information in various formats. Programming screencasts hosted on platforms as YouTube one resource has seen growth popularity adoption over past decade. These usually have some metadata title, short description, set tags should describe what main concepts captured video are. Unfortunately, generic do not detailed about code showcased tutorial, API calls or graphical user interface (GUI) elements employed, could lead missing useful tutorials. Having quick overview GUIs used tutorial can be very helpful for looking examples involving specific calls, design applications with GUI mind. aim is easily available developers, propose VID2META, technique automatically extracts Java import statements, class names, method information, elements, screens from videos makes them metadata. VID2META currently designed work Android screencasts. It analyzes frames using combination computer vision, deep learning, optical character recognition, heuristic-based approaches identify needed frame, extract it, present developer. been evaluated an empirical study 70 programming collected YouTube. results revealed accurately detect average accuracy 90%.