ucas-dx / orgline

The complete training and fine-tuning source code will be released after the paper is accepted.

Introduction

OrgLine, the first multitask analysis pipeline for organoids, which leverages a pretrained detector and a vision foundational model to facilitate large-scale automated cultivation and analysis of organoids. OrgLine achieves precise localization, counting, classification, and segmentation of organoids, providing a foundational and efficient analytical support system.

Getting Started

• Environment Setup Configuration

  Automatic environment setup, please be patient.

  conda create -n orgline python=3.10
  conda activate orgline
  conda install git
  git clone https://github.com/ucas-dx/OrgLine.git
  cd OrgLine
  python install_env.py
  conda install conda-forge::vs2015_runtime 

• Quick usage

  cd models & python simple_inference.py

• Jupyter

  You can run the contents of the SimpleOperation.ipynb file, and all processes will be automated.

  from models.simple_inference import OrgAnalysis
  image_folder = 'images'
  org_analysis = OrgAnalysis(image_folder)
  org_analysis.analyze(show_seg=True,show_bboxes=False)