Here, I share some of my notebooks, working with the PyTorch library on Deep Learning tasks (Computer Vision). All the notebooks are in the context of Image Classification and Object Detection. Models are based on CNN and Vision Transformers (Swin Transformer).

• An image classification using a CNN and regularization terms (dropOut, batch normalization), using the MNIST dataset.
  • Data preprocessing steps
  • DataLoaders
  • Building the model
  • Evaluating the model
• PyTorch Lightning
  • Image classification using a CNN and Cats & Dogs dataset.
  • Fine-tuning a certain number of layers of an ImageNet pre-trained model (Transfer Learning).
• Transfer Learning
  • Fine-tuning the whole network using an ImageNet pre-trained model with the Hymenoptera dataset (Ants & Bees).
  • Fine-tuning a certain number of layers using an ImageNet pre-trained model with the Hymenoptera dataset (Ants & Bees).
• Extracting features of images using VGG16 and using Logistic Regression for the classification stage.