NOAA IFCB Technical Memo Book

Author: Gulce Kurtay, Valentina Staneva (2025)

This book includes Jupyter notebooks that align with the NOAA IFCB Technical Memo. It documents workflows for managing, processing, and analyzing IFCB image data.

Contents

• Data Access and Storage
• Regional Classifier Development through Transfer Learning
• Deploying and Monitoring Classifiers
• Segmentation and Trait Extraction

How to Use This Book

• Browse the chapters on the left sidebar.

• Build the book locally by running:

  jupyter-book build .
  

---

Setting Up Your Environments

This project uses multiple conda environments to run different parts of the workflow.

IFCB Predict Environment (For Deploying and Monitoring Classifiers)

To create and activate the ifcb_predict environment, run:

conda create -n ifcb_predict python=3.10.12
conda activate ifcb_predict

Then, navigate to the folder containing the requirements file and install the dependencies:

cd "C:\Desktop\IFCB-image-data-process\environments"
pip install -r ifcb_predict2.txt

PyTorch Environment (For Segmentation and Trait Extraction)

To create and activate the PyTorch GPU-enabled environment, run:

conda env create -f environments/environment-pytorch.yml
conda activate ifcb-gpu-env