One of the grand visions of AI is the ability to write a computer program from a textual description. This problem is often referred to as code synthesis. Beltramelli and their co-authors (https://arxiv.org/pdf/1705.07962.pdf) recently proposed a system called pix2code, which attempts to solve this problem in some capacity. The goal for this system is to take a Graphical User Interface (GUI) screenshot as its input and generate Domain Specific Language (DSL) code, which can be further compiled as a source code. The following figure, A pix2code Example: A GUI Snapshot on the left is transformed to a simple DSL on the right, shows an example of such an input screenshot on the left and corresponding DSL on the right:
A pix2code ...
