Skip to content

Latest commit

 

History

History
38 lines (27 loc) · 1.08 KB

README.md

File metadata and controls

38 lines (27 loc) · 1.08 KB

Quick3D

Installation

  1. Downloading Pretrained Models

    git clone https://github.com/Daniil-Osokin/lightweight-human-pose-estimation.pytorch.git 
    cd lightweight-human-pose-estimation.pytorch 
    wget https://download.01.org/opencv/openvino_training_extensions/models/human_pose_estimation/checkpoint_iter_370000.pth
    cd ../src/scripts
    !sh download_trained_model.sh
  2. Reach out to me for the configuration files of firebase project if needed! ;)

  3. Setup your own paths where needed.

  4. Run Server

    python server.py
  5. Run localhost or any server you want to test.

  6. open upload.html in your browser. upload interface

  7. Click on Upload button and upload your image and wait for the server to show you 3D model of the corresponding image. viewer

  8. You can view/rotate/zoom/add a texture mesh on the model. rotate

  9. You can also export the 3D model to any 3D formats. export

Android App

We also have the Android app for the 3D Human Digitization.