SALK_projects

Purpose and background

Purpose: segmenting cylinder images and extracting traits from segmentation results for Arabidopsis, rice, soybean, and sorghum.

Background: to be updated with manuscript.

Installation

1. Clone the repository:
   Clone the repository to the local drive.

   git clone https://github.com/Salk-Harnessing-Plants-Initiative/cylinder-penetration.git
   

2. Navigate to the cloned directory:

   cd cylinder-penetration
   

3. Create new conda environments:

   conda env create -f environment.yml
   

4. Activate conda environment

   conda activate segmentation-analysis
   

Running the pipeline

1. crop and segment crop images:

   python pipeline_crop_segment.py --image_path ../Images_test_v03 --save_path ../Segmentation_v03_test  --model_name best_model_crop_cylinder_unetpp_resnet101_1024patch_1batch_40epoch_02_27
   

   Models can be downloaded from Box.

   Change the image_path (../Images_test_v03) to your folder name where you save the cylinder images;

   (optional) Change the save_path (../Images_test_v03) to a folder name where you'd like to save the cropped images and segmentation. The new folder will be created automatically, you don't have to create a new one by yourself.

   Change the model_name (best_model_crop_cylinder_unetpp_resnet101_1024patch_1batch_40epoch_02_27) if needed. Arabidopsis model is best_model_unet_plusplus_resnet101_cylinder_0124; rice model is best_model_rice_seminal_cylinder_unetpp_resnet101_1024patch_4batch_100epoch_05_23; soybean and sorghum model is best_model_crop_cylinder_unetpp_resnet101_1024patch_1batch_40epoch_02_27.

2. get traits and remove outlier:

   python pipeline_analysis.py --image_folder ../Segmentation_v03_test/crop --seg_folder ../Segmentation_v03_test/Segmentation --save_path ../Segmentation_v03_test/analysis --master_data_csv ../MasterData_May2024.csv --plant_group accession
   

   (optional) Change the image_folder (../Segmentation_v03_test/crop) to your folder name where you save the cropped cylinder images in previous step, it is the same one with save_path in previous step plus '/crop' subfolder;

   (optional) Change the seg_folder (../Segmentation_v03_test/Segmentation) to the folder name where you save the segmentation in previous step;

   (optional) Change the save_path (../Segmentation_v03_test/analysis) to the folder name where you'd like to save the original traits and traits after outlier removal;

   Change the master_data_csv (../MasterData_May2024.csv) to the master data file;

   Change the plant_group (accession) to the column in master_data_csv you'd like to remove outlier (based on accession or concentration).