This repository contains the code for figures and analyses presented in Berrigan et al. Fast and efficient root phenotyping via pose estimation.
Before running the analyses, set up the conda environment using the provided env.yaml
file. This ensures that you have all the necessary dependencies installed. Execute the following commands in your terminal:
conda env create -f env.yaml
conda activate berrigan_et_al
Notebook: figure-5/figure_5_error_summary.ipynb
Data:
figure-5/average_root_lengths.csv
figure-5/plantwise_testsets_metrics.csv
Outputs:
- Plots showing median localization error as a percentage of average root length.
- CSV files containing numerical metrics for each model.
Efficient models require fewer labels for accurate predictions and facilitate diverse training sets.
Notebook: figure-8\figure_8_sample_efficiency.ipynb
Data:
-figure-8\sample_efficiency_summary.csv
Outputs:
- Plot showing 90% percentile error vs. number of labeled frame for training per dataset.
- Plot showing number of labeled frames required in training set for model to plateau to within 3 mm of best.
- CSV file of number of labeled frames and 90% percentile error value for each trained model that reached plateau.