BTCCAMP — Python & AI Group Projects

A collection of AI/ML models and teaching materials from BTCCAMP, a summer camp for refugee students held at James Madison University (JMU). I helped TA the Python & AI group, where we introduced students to machine learning fundamentals using hands-on image classification projects.

---

About the Camp

BTCCAMP is a local initiative that teaches programming and technology skills to refugee students in the local area. The Python & AI group covered core concepts including python fundamentals, data preprocessing, model training, evaluation, and real-world applications of machine learning using TensorFlow and Keras.

---

Repository Structure

├── cats&dogs_filtered/     # Training data for lesson (day) 2
├── my_photos/# test data/images
├── myprojects/
│   ├── my_tests/           # Training data and lesson images
│   ├── p1/                 # Dangerous vs. safe animal classifier
│   ├── p2/                 # Fish vs. mammal classifier
│   ├── p4/                 # Clothing type classifier (AI stylist)
│   └── p5/                 # Handwritten digit recognizer
├── Day2A.py                # Lesson code for Day 2 partA
├── Day2B.py                # Lesson code for Day 2 partB
├── Day3.py                 # Lesson code for Day 3
├── python_basics.ipynb     # Example lesson notebook (Day 1)
└── requirements.txt

---

Projects

p1 — Safari (Dangerous) Animal Classifier

A binary image classifier that determines whether an animal is dangerous or safe. Trained on a labeled dataset of animal images using a convolutional neural network (CNN) built with TensorFlow/Keras.

p2 — Deep Sea (Fish vs. Mammal) Classifier

A binary classifier that distinguishes between fish and mammals from images. Demonstrates fundamental binary classification with CNNs.

p4 — AI Stylist (Clothing Classifier)

An image classifier that identifies and differentiates between types of clothing. Inspired by the classic Fashion-MNIST problem, adapted for our own dataset.

p5 — Digit Recognizer

A classifier that recognizes handwritten digits (0–9). A classic introduction to computer vision and multi-class classification.

---

Teaching Materials

The Jupyter notebooks outside of myprojects/ are example lesson files used during the camp:

• Day 1 notebook — Introductory Python and ML concepts covered in class
• Day 2 AI ppxt Introductory to AI and ML concepts covered in class

These were designed to be beginner-friendly for students with little to no prior coding experience.

---

Tech Stack

Tool	Purpose
Python	Primary language
TensorFlow / Keras	Model building and training
NumPy	Numerical operations
OpenCV	Image loading and preprocessing
Matplotlib	Visualizing training results
scikit-learn	Evaluation metrics
Pillow	Image manipulation
Jupyter Notebook	Lesson delivery and prototyping

---

Setup

1. Clone the repo

   git clone https://github.com/Nardobe/BTC-Camp-Python-AI.git
   cd BTC-Camp-Python-AI

2. Create and activate a virtual environment

   python -m venv .venv
   source .venv/bin/activate      # Mac/Linux
   .venv\Scripts\activate         # Windows

3. Install dependencies

   pip install -r requirements.txt

4. Open any notebook

   jupyter notebook

---

Notes

• Models were trained for educational purposes on small datasets — accuracy is intentionally modest to keep training fast and accessible for a camp setting
• p3 is intentionally absent from this repo
• All student-facing materials were designed for beginners with no prior programming experience. Most notebooks are absent for integrity reasons

---

Built and TA'd at JMU · BTC Summer Camp · Python & AI Group