KinetiqVision
← Back to curriculum

Module 3 — Deep learning for vision

Welcome to Module 3

When classical CV ends and learning begins, PyTorch setup, GPU vs CPU, and how Module 3 connects to Modules 1–2.

~25 min read + exercises

Welcome to Module 3 — deep learning for vision

Before we begin

Modules 1–2 were classical CV: hand-designed filters and geometry. Module 3 learns representations from data with convolutional neural networks.

You will train models in PyTorch, practice augmentation and transfer learning, and fine-tune an image classifier.

Figure

Module 3 at a glance

Module 3 — deep vision pathWork top to bottom. Each lesson builds on the previous one.1Welcomeyou are here2CNNsconv nets3Augmenttrain tricks4Transferfine-tune5Quizself-check6Projectclassifier
CNNs, augmentation, transfer learning, quiz, classifier project.

Install before the project

  • Python 3.10+
  • pip install torch torchvision matplotlib numpy
  • Optional GPU: CUDA-enabled PyTorch from pytorch.org

Connection to Modules 1–2

Classical ideaDeep learning version
Convolution filtersLearned conv kernels
Hand-crafted featuresLearned feature maps
RANSAC geometryOften still used with learned detectors

What's next

Lesson 1 — Convolutional networks for images