This repository contains my implementation of a basic image classification project using Convolutional Neural Networks (CNNs) and transfer learning techniques. It serves as a foundational exploration of deep learning for image-related tasks.
The goal of this project is to demonstrate the application of CNNs for classifying images, utilizing well-known architectures and transfer learning methodologies. This approach allows for efficient training on smaller datasets by leveraging pre-trained models.
- Convolutional Neural Networks: Implementation of CNN architectures for effective feature extraction and classification.
- Transfer Learning: Utilizes pre-trained models (e.g., VGG16, ResNet) to improve performance on new datasets with limited samples.
- Data Augmentation: Applied techniques to enhance the dataset and improve model generalization.
- Metrics Monitoring: Integrated with Weights & Biases (WandB) for tracking training metrics and visualizations.
- Framework: PyTorch
- Libraries: NumPy, Matplotlib, WandB
- Datasets: MNIST, Cifar10, Cifar100, FashionMNIST, ASL_Alphabet
Make sure you have the following installed:
- Python
- PyTorch
- WandB