Deep Learning (DL) · Beginner level

Beginner level · Module 2 of 5 · 11 guided hours

Convolutional Networks for Images

Represent images as numbers, understand convolutions and pooling, and build and train a convolutional image classifier.

Learning objectives

  • Represent images as tensors and explain convolution
  • Describe the layers of a convolutional neural network
  • Train and evaluate a simple CNN classifier

Topics

  • pixels, channels and tensors
  • convolution filters
  • feature maps
  • pooling
  • CNN architecture
  • training an image classifier

Lessons

  1. Lesson 1 Images and Convolutions

    See images as grids of numbers and understand how a convolution filter detects patterns such as edges.

    pixel and channeltensor shapeconvolution filterfeature map

  2. Lesson 2 CNN Architecture

    Combine convolution, activation and pooling layers into a network that recognises images.

    convolutional layerpoolingstride and paddingflatten and classifier head

  3. Lesson 3 Training an Image Classifier

    Train and evaluate a CNN on a real image data set and inspect its mistakes.

    training and validation loopsaccuracy and confusion matrixmisclassified examplessaving a model

Practical activity

Train a small CNN on a provided image data set of handwritten digits or simple objects and compare it with a fully connected network.

Expected outcomes

  • Understanding of how CNNs see images
  • A trained, evaluated image classifier

Assessment

Practical: train a CNN to a target accuracy and explain what its first-layer filters respond to.

Free taster lesson Convolutional Neural Networks (CNNs) for Computer Vision Read a free self-study lesson on this topic in the Free Course Library.

Taught live by an instructor — online or in person. Free course lessons are for self-study and do not lead to a certificate — see registration options.