NOBLE Research: Student Guide

This guide walks you through everything you need to work on the project. Bookmark this page. Come back whenever you’re stuck.

Tigris McCauley: Your guide is here: Tigris’s Research Guide

Repo Layout

Here’s what each folder in the repo is for:

Folder	What’s Inside	Do You Touch It?
`notebooks/portfolios/`	Your notebook. This is where you do all your work.	Yes — this is your main workspace
`configs/`	Project settings (image counts, categories, degradation levels)	No — just runs in the background
`src/`	Python scripts that power the notebook (download, dedup, etc.)	No — the notebook calls these for you
`docs/`	This guide	No
`data/`	Where downloaded and generated images are stored locally	Created automatically when you run the notebook
`results/`	Where evaluation metrics and figures will go (later in the semester)	Not yet

You only need to work in notebooks/portfolios/. Everything else supports it.

1. Clone the Repo

Open a terminal (Mac: Terminal app, Windows: PowerShell or Git Bash) and run:

git clone https://github.com/ashleyscruse/ai-generated-image-detection.git
cd ai-generated-image-detection

Don’t have Git? Install it from git-scm.com/downloads.

2. Set Up Your Environment

Create a virtual environment

python3 -m venv venv

Activate it

Mac/Linux:

source venv/bin/activate

Windows:

venv\Scripts\activate

You should see (venv) at the beginning of your terminal prompt. This means it’s active.

Install dependencies

pip install -r requirements.txt

Register the Jupyter kernel

pip install ipykernel
python -m ipykernel install --user --name aidetect --display-name "AI Detect (NOBLE)"

Open your notebook

Open VS Code, then open your notebook at:

notebooks/portfolios/your_name.ipynb

When it asks you to select a kernel, choose “AI Detect (NOBLE)”.

3. Working with Git

You’ll be working on your own branch. This keeps your work separate from everyone else’s.

Switch to your branch

Student	Branch Name
Tigris McCauley	`tigris`
Roland Kindell	`roland`
Brandyn Daley	`brandyn`
Nathan Bryant	`nathan`
Michael Owusu	`michael`

git checkout your-branch-name

Pull the latest changes

Before you start working each time, pull any updates:

git pull origin main

Save your work (commit + push)

After making changes to your notebook:

# Stage your changes
git add notebooks/portfolios/your_name.ipynb
git add data/prompts/

# Commit with a message about what you did
git commit -m "Added 5 prompts and exploration notes"

# Push to YOUR branch
git push origin your-branch-name

Rules

Always push to your own branch, never to main
Pull from main to get updates from me
Commit often – don’t wait until everything is done
Write short, clear commit messages

If you get a merge conflict

Don’t panic. Message me and I’ll help you resolve it.

4. How to Use Your Notebook

Your notebook has 8 parts. Here’s what to do with each:

Part	What’s Inside	What You Do
Part 0: Setup	Imports and paths	Just run it
Part 1: Explore	Browse Open Images dataset	Run it, then answer the questions
Part 2: Download	Download real images for your category	Run it, then answer the questions
Part 3: Prompts	Example prompts + space for yours	Study examples, then write your own
Part 4: Generate	AI image generation (needs GPU)	Run on TACC when ready
Part 5: Degrade	Apply quality degradation	Run it, then answer the questions
Part 6: Readings	Space for reading summaries	Fill in your assigned readings
Part 7: Progress	Progress tracker	Run it to see where you stand

Two types of cells:

Code cells – click the play button to run them. The code works; you don’t need to change it.
“YOUR TURN” sections – markdown cells where you write observations, prompts, and reading notes. Double-click to edit, then click away to save.

5. Storing Images

Images are not stored in Git. They are too large. The .gitignore file prevents them from being committed.

When you download or generate images, they go to:

data/
  raw/
    real/         # Downloaded real images (your category folder)
    synthetic/    # Generated images (your category folder)
  processed/
    clean/        # Original quality
    moderate/     # Surveillance quality
    heavy/        # Bad bodycam quality

These folders exist on your machine (and on TACC) but are not uploaded to GitHub. This is normal.

What IS tracked in Git:

Your notebook (notebooks/portfolios/your_name.ipynb)
Your prompts (data/prompts/your_prompts.json)
Manifests (logs of what you downloaded)

What is NOT tracked:

Image files (.jpg, .png)
Model weights
Large CSV files

If you switch machines (laptop to TACC), you’ll need to re-download images on the new machine. That’s expected.

6. Accessing TACC

We use the Texas Advanced Computing Center (TACC) for GPU-heavy work like generating synthetic images. me will add you to the allocation.

Step 1: Create Your TACC Account

You’ll receive an email from TACC with the subject:

“TACC Project Invitation Action Required: Account Request”

Click the link in the email to create your account
Use your institutional email (e.g., @morehouse.edu, @cau.edu)
Choose a username you’ll remember

Important: Use the link in the invitation email. Don’t create an account separately.

Step 2: Set Up Multi-Factor Authentication (MFA)

TACC requires MFA. Set this up before trying to log in.

Do NOT use SMS/text for MFA. Use an authenticator app.

Go to accounts.tacc.utexas.edu
Click Multi-factor Auth in the sidebar
Set up one of these apps:
- Okta Verify (recommended) – iOS / Android
- Duo Mobile – iOS / Android

Step 3: Log In via SSH

ssh your_username@vista.tacc.utexas.edu

When prompted:

Enter your TACC password
Enter your MFA token (from the authenticator app)

Nothing appears on screen when you type your password. That’s normal. Just type and press Enter.

Windows users: Use PuTTY or Windows PowerShell.

Step 4: Using the TACC Analysis Portal (Jupyter)

For this project, you can also use TACC’s web-based Jupyter interface instead of SSH:

Go to tap.tacc.utexas.edu
Log in with your TACC credentials
Start a Jupyter session on Vista
Upload or clone the repo there
Open your notebook and run the GPU cells (Part 4)

TACC File System

Location	What to put there	Size limit
`$HOME`	Config files, scripts	~10 GB
`$WORK`	Code, datasets, this repo	~1 TB
`$SCRATCH`	Job output, temp files	Unlimited (purged after 10 days)

Clone the repo into $WORK:

cd $WORK
git clone https://github.com/ashleyscruse/ai-generated-image-detection.git
cd ai-generated-image-detection

TACC Checklist

Received TACC invitation email
Created account using the invitation link
MFA set up and tested
Successfully logged in (SSH or Analysis Portal)

7. Getting Help

Stuck for more than 15 minutes? Message me. Don’t spin your wheels.
Git issues? Screenshot the error and send it.
TACC problems? Check the TACC documentation or submit a support ticket.

Quick Reference

Task	Command
Activate venv	`source venv/bin/activate`
Pull updates	`git pull origin main`
Switch to your branch	`git checkout your-branch-name`
Stage changes	`git add notebooks/portfolios/your_name.ipynb`
Commit	`git commit -m "your message"`
Push	`git push origin your-branch-name`
Check branch	`git branch`
SSH to TACC	`ssh user@vista.tacc.utexas.edu`