Skip to content

Latest commit

 

History

History
59 lines (43 loc) · 931 Bytes

README.md

File metadata and controls

59 lines (43 loc) · 931 Bytes

pseudo-labeler

Use EKS as a pseudo-labeler to accelerate pose estimation projects..

Installation

Install ffmpeg

First, check to see if you have ffmpeg installed by typing the following in the terminal:

ffmpeg -version

If not, install:

sudo apt install ffmpeg

Set up a conda environment

conda create --yes --name labeler python=3.8
conda activate labeler

Install dependencies

Lightning Pose:

git clone https://github.com/danbider/lightning-pose.git
cd lightning-pose
pip install -e .
cd ..

Ensemble Kalman Smoother:

git clone https://github.com/paninski-lab/eks.git
cd eks
pip install -e .
cd ..

Install pseudo-labeler package locally

git clone https://github.com/paninski-lab/keypoint-pseudo-labeler.git
cd keypoint-pseudo-labeler
pip install -e .
cd ..

Use

python pipelines/run_pipeline.py --config <path_to_pipeline_config>