SkepticRaven
commented
5 days ago Cross linking the mirrored issue (from the other repo): https://github.com/KumarLabJax/deep-hrnet-mouse/issues/5
@jacobbeierle has been working on a protocols paper provide step-by-step instructions for the video to gait part of the pipeline. I also provided some brief instructions here: https://github.com/KumarLabJax/deep-hrnet-mouse/issues/3 Our gait pipeline should produce an h5 file containing the calculated gait features. If you wanted to inspect them with as a csv, you can use https://github.com/KumarLabJax/gaitanalysis/blob/main/summarizegaitcsv.py to convert one speed bin to a csv. @gautam-sabnis should know the exact speed bin we used in the paper (cursory search didn't list it, was likely 15-20cm/s).
What specifically are you trying to do? In the frailty paper we share the feature data for modeling here: https://zenodo.org/records/6628931 (or more specifically, here: https://github.com/KumarLabJax/vFI-modeling/tree/main/Data), so you should be able to reproduce the the modelling done in the paper.
I'm a graduate student from National Taipei University of Technology in Taiwan. Recently, I read your paper, "A machine-vision-based frailty index for mice", and I am attempting to reproduce the results. However, after setting up the environment and downloading the data from Zenodo and GitHub, I am having difficulty integrating them.
Specifically, when I try to use the tool from "deep-hrnet-mouse" and "gaitanalysis" those two github to run the data from "Stride-level analysis of mouse open field behavior using deep learning-based pose estimation" this zenodo. They do not seem to work together.
For this reason, I am writing to ask for your guidance on how to correctly use these tools and datasets. Can you provide me with the complete and detailed steps to reproduce the results of this paper?