phytooracle/heatmap_ml_based_cropping repository overview

⁠3D Heatmap Detection

This repository contains code for cropping crop central plants using ML on heatmap images from point clouds.

⁠Inputs

Geocorrected 3D point clouds that have undergone pre- and post-processing in the following way: plant names area checked for special characters and empty white space. These characters are removed to prevent any downstream errors.

⁠Outputs

Subdirectories named after individual plants names. Each subdirectory contains multiple 3D point clouds tiles that consitute an individual plant.

⁠Arguments and Flags

• Required Arguments:
  • Path to the point cloud PLY file or directory for batch processing: '-p', '--pcd'
  • Output directory for point cloud: '-po', '--pointcloud_outdir'
  • Output directory for figures: '-fo', '--figures_outdir'
  • Path to model file: '-m', '--model'
  • Pointcloud output filename: '-pf', '--pointcloud_filename'
  • Resolution of the image (meter/pixel): '-r', '--resolution'
  • Confidence threshold for valid plant detections: '-c', '--conf_threshold'