Welcome to my portfolio of deep learning projects, A curated collection of deep learning projects implemented using TensorFlow, Keras, and PyTorch. This repository demonstrates practical applications of neural networks in domains such as image classification, generative modeling, and medical diagnostics, emphasizing clean code, reproducibility, and performance evaluation.
Each subfolder within this repository contains an independent deep learning project, complete with source code, dataset details, training instructions, and results visualization.
| Project | Framework | Domain | Key Topics |
|---|---|---|---|
| Fruit & Vegetable Image Classification | TensorFlow/Keras | Image Classification | CNNs, Image Processing |
| Breast Cancer Prediction using Neural Networks | PyTorch | Binary Classification | Feedforward Neural Networks, Medical AI |
| Fashion MNIST Image Classification | TensorFlow/Keras | Image Classification | CNNs, Regularization |
| Face Mask Detection | TensorFlow/Keras | Binary Classification | Real-Time Detection, CNN, OpenCV |
| MNIST GAN – Digit Generation | TensorFlow | Generative Modeling | GANs, Image Synthesis |
| Anime Face Generator (DCGAN) | TensorFlow | Generative Modeling | DCGAN, Image Generation, Anime Faces |
A convolutional neural network (CNN) designed to classify images of fruits and vegetables.
Technologies: TensorFlow, Keras, Python
Dataset: Kaggle - Fruit and Vegetable Image Recognition
- Data preprocessing and augmentation
- CNN architecture with Dropout and MaxPooling
- Real-time prediction support with confidence scores
- Training and validation performance visualization
📁 Folder: Fruits-and-Vegetables-Image-Recognition-Dataset
A binary classification model developed using PyTorch to predict tumor malignancy from the Breast Cancer Wisconsin dataset.
Technologies: PyTorch, Scikit-learn, Matplotlib
Dataset: sklearn.datasets.load_breast_cancer
- FFNN architecture
- Binary cross-entropy loss with Adam optimizer
- Training curves and evaluation metrics
- Available in Jupyter and standalone script formats
📁 Folder: breast-cancer-prediction
A CNN-based image classifier trained on Fashion MNIST dataset, classifying clothing items into 10 categories.
Technologies: TensorFlow, Keras
Dataset: tensorflow.keras.datasets.fashion_mnist
- Batch Normalization and Dropout layers
- EarlyStopping and ModelCheckpoint callbacks
- Accuracy up to 91–93% with tuning
- Stylish metric visualizations
📁 Folder: Fashion-MNIST-Image-Classification
A real-time CNN classifier that detects whether a person is wearing a face mask or not.
Technologies: TensorFlow, Keras, OpenCV, Python
Dataset: Kaggle - Face Mask Dataset
- Binary classification: With Mask 😷 vs Without Mask 😐
- Data augmentation and preprocessing
- Real-time prediction from user-provided images
- Model saved and reloadable (
.h5) - GPU support and training visualizations
📁 Folder: face-mask-detection
A Generative Adversarial Network (GAN) that synthesizes realistic handwritten digits from the MNIST dataset.
Technologies: TensorFlow, Python
Dataset: tensorflow.keras.datasets.mnist
- Fully functional GAN (Generator + Discriminator)
- Saves generated digit images every epoch
- GAN-stabilization tricks: label smoothing, custom beta values
- Available in both
.pyand.ipynbformats
📁 Folder: mnist-gan
A Deep Convolutional GAN (DCGAN) trained on anime character faces to generate high-quality synthetic images.
Technologies: TensorFlow, Keras, Python
Dataset: Kaggle - Anime Face Dataset
- DCGAN architecture with Conv2DTranspose and LeakyReLU
- Trained on 64×64 anime face images
- Outputs generated image grids every 50 epochs
- Modular code: available in both notebook and script formats
- Excellent visual results for anime face synthesis
📁 Folder: anime-gan
- Clone the Repository
git clone https://github.com/MoustafaMohamed01/DL-Projects.git
cd DL-Projects- Install Dependencies
Each project includes a
requirements.txt. To install dependencies:
pip install -r requirements.txt- Run Projects Navigate to the relevant folder and follow its README to train or run inference.
Contributions are welcome! If you’d like to improve a project or add a new one:
- Fork the repository
- Create a new branch
- Submit a pull request
Ideas, feedback, and improvements are always appreciated.
- LinkedIn: Moustafa Mohamed
- GitHub: MoustafaMohamed01
- Kaggle: moustafamohamed01
- Portfolio: moustafamohamed.netlify.app