Some time ago, I enrolled in the DeepDrive image classification training program [PL] – a great course covering a wide range of topics around image classification, including:

• Best practices for model training in PyTorch and PyTorch Lightning
• Transfer learning and finetuning
• Regularization and data augmentation techniques
• Working with an imbalanced dataset
• Interpretability and explainability
• Self-supervised learning
• Model optimization
• Model deployment

The course offered a fair dose of theoretical background, but the main focus was on practical aspects – each chapter ended with a homework project. I worked through all these projects some time ago, but only recently, I decided to slightly clean up the notebooks, review the results and make all the projects public.

Check them out on my GitHub Repository

PS: If I had to choose only one project to explore, I would pick chapter 08 – interpretability analysis.