Dynamic Web with Automatic Code Generation Using Deep Learning

The typical task assigned to a developer for any design created by a designer is very crucial when it comes to building any software. The process involves tedious and repetitive steps of adapting to rapidly changing client requirements and thus, accommodating those changes in the prototype. This poses a barrier to the software development process. As a solution to this problem, we introduce a model that can be used to automate the process of generating front-end code from hand-drawn wireframes by leveraging deep learning techniques. The image features of the GUI mock-up which is fed to the model as input are extracted to generate markup tags as tokens and rendered in the form of reusable components achieving an overall BLEU score of 0.92.