Full Convolutional Front-End Code Generation Based on Web Page Screenshot
Download as PDF
DOI: 10.23977/CNCI2020047
Author(s)
Zihao Chen and Xiao Li
Corresponding Author
Zihao Chen
ABSTRACT
Nowadays, Internet has gained its popularity, so more and more kinds of web applications need to be developed. For that, software company require a generator to output front-end code immediately based on web page screenshot to Improve development efficiency. Using full convolutional algorithm which contains CAE and CNN, we proposed a method performs better than pix2code. Our method follows the encoder-decoder framework and add hierarchical attention mechanism into it which learn from image caption methods. We use the PixCo dataset to complete the experiment.
KEYWORDS
Code generation; image caption; CNN; front-end development; DSL