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

YOLaT & YOLaT++

YOLaT & YOLaT++ are graph-based machine-learning models for recognizing vector graphics from textual vector representations without rasterizing them. Their public repository contains the official PyTorch implementation, Python scripts for preparing floorplan and diagram datasets, and training commands. YOLaT++ models vector graphics hierarchically across primitives, curves, and points, with position-aware enhancement

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

YOLaT & YOLaT++ — 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 the repository when developing or studying recognition workflows that can operate on vector structure and when you can set up the documented Python environment and datasets.

Who it is for

  • Researchers studying vector-graphics recognition.
  • Developers building floorplan or diagram recognition workflows from vector data.

Documented uses

  • Prepare floorplan or diagram datasets as graphs for model workflows.
  • Train graph-based recognition models on the documented datasets.
  • Explore hierarchical recognition across vector primitives, curves, and points.
Before you choose

Requirements & limits

The repository is public and specifies an MIT license. Its README documents a Python 3.8 Conda environment and an installation script.

  • The documented workflows use Python 3.8, Conda, PyTorch, and repository-provided installation and data-preparation scripts.
  • The README gives dataset-specific preparation and training instructions for floorplans and diagrams.
  • The project provides code and scripts rather than a hosted recognition application.
  • YOLaT++ adds a three-level hierarchy of primitives, curves, and points and position-aware enhancement.

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

Prepare floorplan graphs for training or inference

Convert the documented floorplan dataset into graph data using the repository’s preparation script.

Before you start

  • Python 3.8 Conda environment and the repository installation script.
  • The Floorplans dataset downloaded and unzipped in data/FloorPlansGraph5_iter.

Steps

  1. Create and activate a Conda environment with Python 3.8.

  2. Run sh deepgcn_env_install.sh from the repository.

  3. Place the unzipped Floorplans dataset in data/FloorPlansGraph5_iter.

  4. From the repository’s utils directory, run python svg_utils/build_graph_bbox.py.

Expected outcome

The repository’s documented floorplan graph-preparation stage is run for training or inference data.

Guide limitations

  • This guide covers dataset preparation only; it does not run training or inference.
  • The README does not detail dataset contents or troubleshooting.
Documentation 1 ↗
Source-based practical guide

Prepare diagram graphs for training or inference

Convert the documented diagram dataset into graph data using the repository’s preparation script.

Before you start

  • Python 3.8 Conda environment and the repository installation script.
  • The Diagrams dataset downloaded and unzipped in data/diagrams.

Steps

  1. Create and activate a Conda environment with Python 3.8.

  2. Run sh deepgcn_env_install.sh from the repository.

  3. Place the unzipped Diagrams dataset in data/diagrams.

  4. From the repository’s utils directory, run python svg_utils/build_graph_bbox_diagram.py.

Expected outcome

The repository’s documented diagram graph-preparation stage is run for training or inference data.

Guide limitations

  • This guide covers dataset preparation only; it does not run training or inference.
  • The README does not detail dataset contents or troubleshooting.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
17 October 2026
Original website title
microsoft/YOLaT-VectorGraphicsRecognition · Official homepage title · retained in its source language
How we review sourcesSuggest a correction