CS7002 - AI Vision and Deep Learning (2024/25)
Module specification | Module approved to run in 2024/25 | ||||||||||||
Module title | AI Vision and Deep Learning | ||||||||||||
Module level | Masters (07) | ||||||||||||
Credit rating for module | 20 | ||||||||||||
School | School of Computing and Digital Media | ||||||||||||
Total study hours | 200 | ||||||||||||
|
|||||||||||||
Assessment components |
|
||||||||||||
Running in 2024/25(Please note that module timeslots are subject to change) |
|
Module summary
The module is designed to impart essential mathematical principles and concepts of computer vision alongside its practical applications. The module encompasses core topics in image formation and low-level image processing; mid-level scene representation; model-based description and tracking. Appropriate hardware/software tools will be integrated into the module to enable students to apply and test computer vision algorithms including deep learning on real world data sets.
The objectives of this module are to:
Enable students to gain understanding of essential concepts of image processing / computer vision algorithms with their practical applicability in real systems.
Develop students’ expertise in analysing, designing, building, training, and evaluating computer vision deep learning models.
Train students in using appropriate hardware / software tools for solving common computer vision problems.
Prepare students with postgraduate level research and report writing skills.
Syllabus
Introduction to Computer Vision (CV) and basic concepts: image formation, light, colour and human visual perception, geometric transformation, geometric camera models, review of basic mathematics, high level insight of CV applications and applications [LO1]
Pre- and post- image processing: image representation, pixel layouts, point and neighbourhood operations, filtering, histogram, transforms, mathematical morphology, segmentation [LO1, LO2, LO3]
Visual features: edge, corner, colour, texture, boundary and shape, scale invariant feature transform (SIFT) [LO1, LO2, LO3]
Recognition: introduction to pattern recognition, decision function and elementary statistical decision theory, classification problems, object detection, region-of-interest estimation, clustering, dimensionality reduction, template matching, non-maximum suppression (NMS) [LO1, LO2]
Advanced concepts, applications, and challenges: Deep Learning (DL), Multilayer perceptron (MLP), Convolutional Neural Networks (CNN), Single-Shot Detectors (SSD), advanced CNN architectures, training datasets, transfer learning, activation functions, feedforward process, error functions, backpropagation, regularization techniques, real world problems and limitations of CV [LO4].
Visual data analytics issues: legal, social, ethical, and professional (LSEP) issues related to image processing and visual data analytical systems [LO5].
Balance of independent study and scheduled teaching activity
The module will be delivered by a mixture of lectures, tutorials / workshops, synchronous / asynchronous e-Learning / blended learning. Students will be expected to carry out directed independent background study to familiarise themselves with the platforms and tools that will be used during the module.
All required learning material for the module (lectures / tutorials / workshops / recordings / links etc.) will be available to students on University VLE (WebLearn). The module page on VLE will be continuously updated with announcements and need based additional information to ensure appropriate support for students’ learning.
Students will be encouraged to keep logbook for reflective learning and regular feedback.
Learning outcomes
At the end of this module the student should be able to:
LO1. Understand and apply underpinning mathematics and / or physics governing computer vision algorithms / systems.
LO2. Demonstrate sound understating of the theory and operation of image processing and computer vision algorithms / systems, and a critical awareness of current problems and new insights.
LO3. Use software / hardware and modelling tools to analyse and implement selected aspects of computer vison algorithms / systems.
LO4. Develop postgraduate level skills in literature review, critical evaluation of results and report writing by exploring advanced topics and / or recent related to computer vision algorithms / systems. Build intuition behind structuring computer vision Deep Learning projects and hyperparameters tuning.
LO5. Show awareness of legal, social, ethical, and professional (LSEP) issues particularly important in computer vision algorithms and systems.
Bibliography
Core Text:
Mohamed Elgendy (2020). Deep Learning for Vision Systems.
Richard Szeliski (2022). Computer Vision: Algorithms and Applications.
E. R. Davies (2017). Computer Vision: Principles, Algorithms, Applications, Learning, Academic Press. Available at LMU library: https://research.ebsco.com/c/d5khri/search/details/e4crp2kbwf?limiters=FT1%3AY&q=Computer%20Vision%3A%20Algorithms%20and%20Applications
Sonka, Hlavac & Boyle (2015). Image Processing, Analysis and Machine Vision. Available at LMU library: http://catalogue.londonmet.ac.uk/record=b1677110~S1
Gonzalez et al (2008). Digital Image Processing. Available at LMU library: http://catalogue.londonmet.ac.uk/record=b1556623~S1
Other Reading:
Christopher M. Bishop (2006). Pattern recognition and machine learning. Available at LMU library: http://catalogue.londonmet.ac.uk/record=b1576490~S1
Hafsa Asad, Vishwesh Ravi Shrimali, et al. (2020). The Computer Vision Workshop: Develop the skills you need to use computer vision algorithms, Packt Publishing.
Wesley E. Snyder, Hairong Qi (2010). Machine Vision, Cambridge University Press. Available at LMU library: http://catalogue.londonmet.ac.uk/record=b1765736~S1
Jonathan Fernandes (2019). Introduction to Deep Learning with OpenCV, AV Presentation from LinkedIn Learning. Available at LMU library: http://catalogue.londonmet.ac.uk/record=b2268222~S1
Joseph Howse, Joe Minichino (2020). Learning OpenCV 4 computer vision with Python 3: get to grips with tools, techniques, and algorithms for computer vision and machine learning.
Prateek Joshi (2015). OpenCV with Python by example: build real-world computer vision applications and develop cool demos using OpenCV for Python.
Research Databases:
IEEE Xplore: http://catalogue.londonmet.ac.uk/record=b1572993~S1
ACM Digital Library: http://catalogue.londonmet.ac.uk/record=b1251310~S1
Wiley Online Library: http://catalogue.londonmet.ac.uk/record=b1585296~S1
London Met Library: https://student.londonmet.ac.uk/library/
Journals:
IET Computer Vision: https://ietresearch.onlinelibrary.wiley.com/
IEEE Pattern Analysis and Machine Intelligence
IEEE Transactions on Image Processing,
IEEE Transactions on Computational Imagining
ELSEVIER Pattern Recognition
Springer Internal Journal of Computer Vision: http://catalogue.londonmet.ac.uk/record=b2045879~S1
IPSJ Transactions on Computer Vision and Applications: https://ipsjcva.springeropen.com/
Visualization in Engineering: https://viejournal.springeropen.com/
LinkedIn Learning: https://www.linkedin.com/learning/