You can use it to automatically remove the mosaics in images and videos, or add mosaics to them.
    This porject based on "semantic segmentation" and "Image-to-Image Translation".

    More example

    origin auto add mosaic auto clean mosaic
    image image image
    image image image
    mosaic image DeepCreamPy ours
    image image image
    image image image
    • Style Transfer
    origin to Van Gogh to winter
    image image image

    An interesting example:Ricardo Milos to cat

    Run DeepMosaics

    You can either run DeepMosaics via pre-built binary package or from source.

    Pre-built binary package

    For windows, we bulid a GUI version for easy test.
    Download this version and pre-trained model via [Google Drive] [百度云,提取码1x0a]


    • Require Windows_x86_64, Windows10 is better.
    • Different pre-trained models are suitable for different effects.[Introduction to pre-trained models]
    • Run time depends on computer performance(The current version does not support gpu, if you need to use gpu please run source).
    • If output video cannot be played, you can try with potplayer.
    • GUI version update slower than source.

    Run from source



    This code depends on opencv-python, torchvision available via pip install.

    Clone this repo

    git clone
    cd DeepMosaics

    Get pre-trained models

    You can download pre_trained models and put them into './pretrained_models'.
    [Google Drive] [百度云,提取码1x0a]
    [Introduction to pre-trained models]

    Simple example

    • Add Mosaic (output media will save in './result')
    python3 --media_path ./imgs/ruoruo.jpg --model_path ./pretrained_models/mosaic/add_face.pth --use_gpu 0
    • Clean Mosaic (output media will save in './result')
    python3 --media_path ./result/ruoruo_add.jpg --model_path ./pretrained_models/mosaic/clean_face_HD.pth --use_gpu 0

    More parameters

    If you want to test other image or video, please refer to this file.

    Training with your own dataset

    If you want to train with your own dataset, please refer to


    This code borrows heavily from [pytorch-CycleGAN-and-pix2pix] [Pytorch-UNet] [pix2pixHD] [BiSeNet].





    贡献者 2

    HypoX64 @weixin_36721459
    H HypoX64 @HypoX64


    • Python 100.0 %