PyTorch Essential Training: Working with Images
1h 2mIntermediate2025-04-04
Authors
Terezija Semenski
Software Developer, Mathematician, Writer, and Learner
Course details
Machine learning developers and data scientists can feel overwhelmed by the complexity of convolutional neural networks (CNNs) and their applications. This course provides a hands-on, project-based approach to image classification. Join instructor Terezija Semenski to gain practical experience in preprocessing data, training, and evaluating a pre-trained model. Plus, explore transfer learning, model fine-tuning, and evaluation metrics.
Learning objectives
Apply a pre-trained deep learning model and to a binary classification task.
Preprocess image data and fine-tune the ResNet model.
Evaluate and improve the performance of the model using accuracy and loss.
Learning objectives
Apply a pre-trained deep learning model and to a binary classification task.
Preprocess image data and fine-tune the ResNet model.
Evaluate and improve the performance of the model using accuracy and loss.
Skills covered
NumPyPyTorchNeural Networks and Deep LearningPythonArtificial Intelligence (AI)Open SourceOne-Off
Concepts
Introduction
- Image classification with PyTorch
- What you should know
Introduction to Image Classification
- What is image classification
- Binary image classification
- Multiclass image classification
- Understanding convolutional neural networks (CNNs)
Data Preparation
- Importing the packages
- Organizing the dataset
- Transforming the data
- Visualizing the data
Transfer Learning with Pretrained Model
- Introduction to transfer learning
- ResNet model
- Fine-tuning ResNet for binary classification
- Freezing layers and training from specific blocks
Training and Testing the Model
- Setting up the training loop
- Loss function and optimizer
- Evaluating model performance
- Saving the model
- Visualizing predictions
- Challenge - Evaluate and test the model
- Solution - Evaluate and test the model
Capstone Project
Conclusion
- Next steps