Skip to content

Human height estimation from video live feed with deep learning based automatic scene calibration

License

Notifications You must be signed in to change notification settings

MohammadDallash/Height-Anything

Repository files navigation

HeightAnyone

HeightAnyone is a framework built on top of the state-of-the-art depth estimation model, DepthAnyThingV2, and YOLOv9 for segmentation, designed for human height estimation, This framework does not require any information about the camera or scene parameters, except for at least one vertical known height in the video.

demo

Usage

1. Install Requirements:

pip install -r requirements.txt

2. Download Models (around 1.3GB)

./download_models.sh

3. Follow along the steps in the pipeline notebook

Minimum Requirements

  • GPU with at least 3GB of VRAM

About

Human height estimation from video live feed with deep learning based automatic scene calibration

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published