Skip to content
Get packs

Animated Drawings

Turns a child's drawing of a human figure into an animated character using a pose-detection pipeline.

GitHub Download v0.0.1
Stars
12.8k
Forks
1.2k
Licence
MIT
Language
Python
Last update
1.2 years ago
Type
AI model / agent

This repository is archived and no longer maintained.

What it is

Animated Drawings is the code that accompanies a research paper on animating children's drawings of people. You give it a drawing, it detects the figure and estimates its joints, then plays a motion clip on that character. The README also presents it as a creative tool for making animations with your own drawn characters. The project is archived, so the maintainer no longer updates it.

  • Animate a drawn character by running image_to_animation.py on a drawing file
  • Export an animation as an MP4 video using an example config
  • Export a transparent GIF of an animated character
  • Render animations headlessly on a remote server by enabling USE_MESA in the config

Great for

  • Developers curious about the paper's method and want to run its code
  • Hobbyists who want to animate simple drawings of people
  • Teachers or students exploring pose estimation with a visual result
  • Researchers who want a reference implementation to build on

Not ideal for

  • Anyone needing ongoing support, since the repository is archived and the maintainer has stopped maintaining it
  • People who need reliable results on non-humanoid or highly stylized drawings, since the models target the human figure
  • Users on operating systems other than those tested (the README names macOS Ventura and Ubuntu 18.04)

How to get it

  1. 1

    Clone and build from source

    You'll need: Miniconda or another Python virtual environment manager · Python 3.8.13 (as specified in the README) · Git · Docker Desktop (optional, for the Docker option)

    Create and activate the conda environment

    conda create --name animated_drawings python=3.8.13

    Activate the environment

    conda activate animated_drawings

    Clone the repository

    git clone https://github.com/facebookresearch/AnimatedDrawings.git

    Install the package from the repo folder

    cd AnimatedDrawings && pip install -e .
  2. 2

    Run it

    • Activate the animated_drawings conda environment and move into the AnimatedDrawings directory
    • Start a Python interpreter and run: from animated_drawings import render
    • Run render.start('./examples/config/mvc/interactive_window_example.yaml') to open an interactive window
    • Use spacebar to pause, arrow keys to move through time, and q to close
  3. 3

    Get the most out of it

    • Edit the YAML config files to change characters, motions, and scenes, as described in examples/config/README.md
    • Use export_mp4_example.yaml or export_gif_example.yaml to save results to files
    • Set USE_MESA: True in the view section of a config for headless rendering
    • If joint predictions are wrong, run fix_annotations.py to open a web interface and correct them
    • Edit mask.png in an image editor if the mask prediction is wrong

Downloads

Source code (.zip)main branch · straight from GitHubRelease v0.0.18 files incl. drawn_humanoid_pose_estimator.mar

Strengths

  • Comes from the authors of a published paper, with the method described in the README
  • Configuration files control characters, motions, and export format without rewriting code
  • Includes both Docker and a local macOS option for the detection service

Trade-offs

  • Archived, so no further updates or bug fixes are expected
  • Setup has several moving parts (Python environment, TorchServe, Docker or a macOS script) and the README warns it can be complicated

Stats synced from GitHub. The assessment above was drafted with AI from the project's README on 2026-10-29 and may miss recent changes — check the README before relying on a command. MIT licence: free to use, modify and distribute, including commercially, as long as you keep the copyright notice.