README.md 889 B
FreiPose
FreiPose: A Deep Learning Framework for Precise Animal Motion Capture in 3D Spaces
We present a general purpose framework to estimate 3D locations of predefined landmarks (f.e. skeletal joints of an animal) given information from multiple cameras using deep learning methods. Our frame work is configurable which makes it easy to adapt to the specific experimentation needs and general in terms of number, type and location of cameras used as well as adaptable over a wide range of subjects of interest (animals, objects, humans, ...). Our goal is to bring deep learning based methods into the labs, without the need of expert knowledge about machine learning.
This repository containes example data sets as well as Source Data from figures.