Speedy Measurement of Arabidopsis Rosette Traits (SMART)

Author: Suxing Liu

Robust and parameter-free plant image segmentation and trait extraction.

1. Process with plant image top view, including whole tray plant image, this tool will segment it into individual images.
2. Robust segmentation based on parameter-free color clustering method.
3. Extract individual plant gemetrical traits, and write output into excel file.

Requirements

Either Docker or Singularity is required to run this project in a Unix environment.

Usage

Docker

docker pull computationalplantscience/smart
docker run -v "$(pwd)":/opt/arabidopsis-rosette-analysis -w /opt/arabidopsis-rosette-analysis computationalplantscience/arabidopsis-rosette-analysis python3 /opt/arabidopsis-rosette-analysis/trait_extract_parallel.py -i input -o output -ft "jpg,png"

Singularity

singularity exec docker://computationalplantscience/arabidopsis-rosette-analysis python3 trait_extract_parallel.py -i input -o output -ft "jpg,png"

Contents

• Requirements
• Usage

Requirements

The easiest way to run this project is with Docker or Singularity .

To pull the computationalplantscience/smart image, the current working directory, and open a shell with Docker:

docker run -it -v $(pwd):/opt/dev -w /opt/dev computationalplantscience/smart bash

Singularity users:

singularity shell docker://computationalplantscience/smart

Usage

Segmentation

To perform color segmentation:

python3 /opt/smart/core/color_seg.py -p /path/to/input/file -r /path/to/output/folder

You can also pass a folder path (-p /path/to/dir). By default any JPG and PNG are included. You can choose filetype explicitly with e.g. -ft jpg.

To extract traits:

python3 /opt/smart/core/trait_extract_parallel_ori.py -p /path/to/input/file -r /path/to/output/folder

You can also use a folder path as above, likewise for filetype specification.

By default this script will not perform leaf segmentation and analysis. To enable leaf analysis, use the -l flag.

To indicate that your input is a multiple-tray or -individual photo, add the -m flag.