2018 · Conference paper

pix2code: Generating Code from a Graphical User Interface Screenshot

Tony Beltramelli

Learns screenshot-to-DSL generation for native Android XML, iOS Storyboards and web HTML/CSS interfaces, then compiles the DSL into platform UI code.

Scope and limitations

Evaluation uses synthetic screenshot/code pairs: 1,500 training and 250 test instances per platform. Reported accuracy concerns DSL tokens, not complete apps or behavior. Text labels are ignored; platform datasets train separate models using the same architecture. The task covers layout, components and relationships, not application logic or backend recovery. The repository describes an educational research prototype; its older dependencies were not runtime-tested.

Why it fits this collection

Adds an early learned screenshot-to-code approach spanning native mobile and web UI reconstruction.

Publication and review details

Venue: EICS 2018
Status: Conference paper
Publication / posting date: June 19, 2018
Date source: ACM-deposited Crossref proceedings metadata: online and print publication on 19 June 2018; arXiv first submitted 22 May 2017.
First arXiv submission: May 22, 2017
arXiv: 1705.07962
Reviewed arXiv version: v2
DOI: 10.1145/3220134.3220135

The published EICS 2018 proceedings article follows the 2017 arXiv release; this record reviews arXiv v2 dated 19 September 2017. The linked author repository is the implementation identified by that version, reviewed at a current pinned commit. It contains compiler code and split dataset archives; the actual LICENSE declares Apache 2.0, while the README license badge has a stale filename.

Sources reviewed 2026-10-16