sketch-code
SketchCode is a Keras-based deep learning model that converts hand-drawn web drafts into HTML code. It uses an image tagging architecture that includes CNN encoder and LSTM decoder to process input images. It is trained with a data set of about 1,700 synthetic visuals, aiming at TensorFlow 1.x and the Keras library of that period.
ashnkumar · Image & Photo Processing
Common commercial products in this category: Adobe Photoshop, Adobe Lightroom, Cloudinary
- Using pre-trained weights, you convert a single drawing image into HTML code.
- You convert multiple drawing images in a folder to HTML format in bulk.
- You can train the model from scratch with your own data set and apply data boost parameters.
- • No published release
- • NeutralGitHub Issue
[Dataset Added] : 100 + images and their lables
“Hi guys , I found 100+ hand-drawn wireframe dataset and some wireframes are drawn by myself How I train model for new wireframe dataset and used them in this project from scratch ? From beginning of installion of packages to generate HTML code Thank you :)”
- • NeutralGitHub Issue
Dataset and Pre-trained Model Files Not Accessible
“Dataset and the pre-trained model files are not accessible. Getting 403 Forbidden from the server `$ sh get_data.sh mkdir: cannot create directory ‘../data’: File exists --2022-03-18 12:14:57-- Resolving sketch-code.s3.amazonaws.com (sketch-code.s3.amazonaws.com)...…”