Skip to content

Clustering of plant identifications over a season for individual plant tracking

Notifications You must be signed in to change notification settings

phytooracle/rgb_plant_clustering

Repository files navigation

RGB Identification Agglomerative Clustering

This repository contains the agglomerative clustering scripts used in the PhytoOracle pipeline to track plants over time for each season, sorted and containerized by branch.

Inputs

The containers take in a csv with at least the columns of 'genotype', 'lat', 'lon'.

Optional Input: An additional csv containing 'genotype', 'lat', 'lon' using the flag '-r'. Whatever clusters contain these points will get a 1 in their 'double_lettuce' column.

Outputs

The containers will return a csv with an added columns 'plant_name' that is in the format (f'genotype',f'match number'), where match number is the order in which the plant was matched in that genotype. eg: The first plant that is clustered in the Franklin genotype will be named (Franklin, 1). There will also be a 'double_lettuce' columns that indicates if the lettuce was clustered with any of the points provided in your 'remove_points' csv.

The outputted csv is easy to use with pandas, but can be difficult to read, so in the repository containing this readme there is also a csv to json conversion script that creates a more human readable document.

Arguments and Flags

  • Positional Arguments:

    • Directory containing CSV files to cluster: 'csv_list'
  • Optional Arguments:

    • Remove Points: '-r', '--remove_points'
    • Output directory: '-o', '--outdir', default='pointmatching_out'
    • Output filename: '-f', '--filename', default='agglomerative_plant_clustering'

About

Clustering of plant identifications over a season for individual plant tracking

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published