Computer Vision

Module aims

In this module you will learn how images are formed, how they are represented on computers and how they can be processed by computers to extract semantic information. As part of the module you will have the opportunity to develop algorithms for detecting interesting features in images, design neural networks to perform natural image classification and explore algorithms for solving real-world problems such as hand-written digit recognition and object detection.   

Learning outcomes

Upon successful completion of this module you will be able to:

  • Describe the process of image formation
  • Design filters for image processing, edge detection and interest point detection
  • Understand features and classifiers in the context of image classification
  • Implement neural networks for image classfication
  • Understand ideas underlying object detection and image segmentation
  • Recall commonly used methods for motion estimation   

Module syllabus

This module covers the following topics:

  • Image formation
  • Image filtering
  • Edge detection and interest point detection
  • Feature descriptors
  • Image classification
  • Object detection and image segmentation
  • Neural networks
  • Motion estimation
     

Teaching methods

The teaching approach is centred around the desire to solve real-world visual information processing problems, such as natural image classification, hand-written digit recognition and object detection. Such examples are used throughout to demonstrate how the principles taught can be applied in practice.

The concepts that you have learnt in lectures will be reinforced by unassessed, formative, tutorial exercises and assessed computer-based courseworks. The courseworks will cover both low-level and high-level image processing. The lab sessions are supervised, so you will receive technical support from Graduate Teaching Assistants (GTAs).

The Piazza Q&A web service will be used as an open online discussion forum for the module.
 

Assessments

The two computer-based courseworks count for 20% of the marks and the exam counts for the remaining 80% of the marks. The courseworks are in the format of jupyter notebook, which enables you to fill in source code, discuss your solutions and display results as a pdf file for submission.   
There will be detailed written feedback for each of the assessed courseworks and class-wide feedback explaining common pitfalls and suggestions for improvement.