Skip to content

Latest commit

 

History

History
58 lines (30 loc) · 1.3 KB

README.md

File metadata and controls

58 lines (30 loc) · 1.3 KB

This repository presents an university project done with a classmate for single/multi person detection in images and videos, and height inference (only on images) on android devices.

The code for the the single person ist based on the tensorflow lite example code (https://github.com/tensorflow/examples/tree/master/lite/examples/posenet/android) and the code of the multiperson detection is based on the tensorflow lite flutter plugin code (https://github.com/shaqian/flutter_tflite/tree/master/android, https://medium.com/flutter-community/posenet-for-ios-android-and-flutter-using-tensorflow-lite-836788a110c7 ).

AndroPose

Human pose estimation for (android) mobile device

1. Pose estimation in image

  • Original image

reiter-original


  • After estimating pose using tflite

reiter-modified



2. PoseEstimation in Video


video


3. New design (AndroPose 2.0): coming soon

The version 3.0 of AndroPose is already available here



design


3.1 Application test on new design


img-detect-new-design


Display image reconstruction


img-reconstruction