OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Models

SketchCode

SketchCode is an open-source Python and Keras model that generates HTML markup from images of hand-drawn website wireframes, using a CNN encoder and LSTM decoder.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

SketchCode — OpenViST project profile in AI Models.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use it to explore or prototype sketch-to-HTML workflows when input wireframes resemble the model's synthetic training examples.

Who it is for

  • Developers prototyping starter HTML from drawn website mockups.
  • People studying image-captioning architectures applied to code generation.

Documented uses

  • Convert a single PNG wireframe into HTML.
  • Process a folder of wireframe images into HTML using the documented batch script.
  • Train the model on GUI images or evaluate predictions using the repository's documented scripts.
Before you choose

Requirements & limits

The public GitHub repository provides the code under an MIT license. The README states that portions derive from pix2code and require the attributions and terms described in its NOTICE file.

  • The model was trained on a 1,700-image synthetic dataset and is not designed to generalize across the variability of actual wireframes.
  • The documented stack targets Python 3, TensorFlow 1.x, and the Keras version of that era; the repository is no longer maintained.
  • The README documents pretrained weights, single-image and batch conversion, training, and BLEU-score evaluation workflows.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Convert one wireframe image into HTML

Use the repository's documented pretrained-model workflow to convert one PNG wireframe into HTML output.

Before you start

  • Python 3 and pip
  • Project dependencies installed with pip install -r requirements.txt
  • Repository data and pretrained weights downloaded with the documented scripts

Steps

  1. Clone the repository and enter its scripts directory.

  2. Run sh get_data.sh and sh get_pretrained_model.sh to retrieve the data and pretrained model files.

  3. From the src directory, run convert_single_image.py with the PNG path, output folder, model JSON file, and model weights file arguments.

Expected outcome

The script is configured to write generated HTML to the chosen output folder.

Guide limitations

  • The project describes itself as a proof of concept; results rely on wireframes resembling its synthetic training data.
  • The repository targets TensorFlow 1.x and the Keras stack of its era and is no longer maintained.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
14 October 2026
Original website title
ashnkumar/sketch-code · Official homepage title · retained in its source language
How we review sourcesSuggest a correction