Computer Vision: CNNs, Detection, Segmentation and Vision Transformers: A University Exam Revision Guide (Full Marks Press: AI Study Guides)
Computer vision, from the pixel to the vision transformer.
Computer Vision is a concise university study guide for the computer vision or image processing module that sits alongside machine learning in most computing degrees. It is written for the exam rather than the reading list: every chapter builds the idea, works an example the way an examiner marks it, and then asks you to do one yourself. Nothing is left as an exercise for the reader at the point it matters.
What this guide coversImage representation, sampling, and digital imagesColour: models, spaces, and perceptionImage preprocessing, enhancement, and normalisationData augmentation and dataset constructionClassical features: edges, corners, and gradientsLocal feature descriptors and matching (SIFT-style)Convolution, filtering, and the frequency domainConvolutional neural networks: the building blocksImage classification architecturesTransfer learning and fine-tuningObject detectionSemantic and instance segmentationModern architectures: vision transformers and beyondEvaluation, benchmarking, and responsible computer visionWho it is for
Undergraduate and taught-masters students sitting a computer vision module, and anyone revising the subject properly rather than skimming it. It also works as a fast, honest catch-up for practitioners and career changers who need the foundations underneath the tools they already use. You need linear algebra and the basics of neural networks. The classical image-processing chapters assume nothing.
How it is written
Worked examples are marked the examiner’s way, so you see not just the answer but why it earns the grade: how a convolution changes an image’s frequency content, and what that means for the filter you chose. Responsible AI sits inside the technical chapters where the decisions are actually made, not in a closing sidebar you can skip. The edition date is printed on the cover, so you can see for yourself how current the material is instead of taking a publisher’s word for it.
Inside
14 chapters, worked examples with the marks shown, practice questions, and diagrams that carry the explanation rather than decorating it. Written by practitioners who build and evaluate these systems, and published by Full Marks Press.
Why the edition date is on the cover
AI moves fast enough that a study guide written three years ago can teach you something an examiner will mark down. Most guides hide their age. This one prints its edition date on the front, states what was current when it was written, and says plainly where the field is still moving. You can judge for yourself whether it is current enough for the paper you are sitting, which is more than a reading list will give you.
How to use it
Read a chapter, work its example on paper before you look at the marking, then answer the practice questions cold. If you can reconstruct the worked example without the book, you know that chapter. If you cannot, you have found exactly what to revise, which is the whole point of revising with a pen rather than a highlighter. The chapters are ordered so each one only assumes what came before it, so you can work straight through or drop into the one topic that is not sticking.
The series
Computer Vision is one of fourteen guides in Full Marks Press: AI Study Guides, one per university module, all to the same standard. If this one is on your timetable, MLOps, Mathematics for AI and Programming for AI probably are too.
Current enough to trust, rigorous enough to earn the grade. Aim for full marks.
From the Publisher
Fourteen exam companions. One per module. One standard.
Practitioner-written, and built for the mark scheme
Full Marks Press publishes concise revision guides for the modules that make up an AI, machine learning or data science degree. One guide per module. Each is written for the exam rather than the reading list: build the idea, work an example the way an examiner marks it, then hand the question over to you. Written by practitioners who build and evaluate these systems. Current enough to trust, rigorous enough to earn the grade.
What is inside every Full Marks Press guide
Worked examples, marked like an exam
Every chapter ends with a worked example annotated the way an examiner annotates it. You see where each mark is earned, not just the final answer, and which slips cost candidates the marks they should have had.
The edition date is on the cover
AI moves fast enough that a guide written three years ago can teach you something an examiner marks down. Most guides hide their age. These print the date on the front and say plainly where the field is still moving.
Responsible AI inside the chapter
Bias, evaluation, privacy and safety are not a closing sidebar you can skip. They sit in the chapters where the decisions are made, because that is where they are examined and that is where they matter.
Diagrams that do the work
Every diagram carries part of the explanation rather than decorating the page. Backpropagation, attention, the Bellman backup: drawn so the mechanism is visible, then worked through in the text beside it.
Price
— no data
$5.99$5.99 $5.99$5.99 $5.99$5.99 $5.99$5.99 $5.99$5.99
Typical module year
Year 1 Year 2 Year 2 to 3 Year 3 or MSc Year 3 or MSc Year 2 to 3
What it covers
Linear algebra, calculus, probability, optimisation Regression, classification, ensembles, evaluation Backpropagation, CNNs, RNNs, training Attention, transformers, LLMs, alignment, prompting Embeddings, transformers, language tasks Filtering, CNNs, detection, segmentation
Maths assumed
A-level maths School algebra and a little calculus Matrices, partial derivatives, chain rule Neural network fundamentals Basic probability, some Python Linear algebra and CNN basics
Read this first
Nothing Mathematics for AI Machine Learning Neural Networks Machine Learning Neural Networks
Pages
148 138 143 160 155 161
Free sample chapter
Yes Yes Yes Yes Yes Yes
Full Marks Press guides from the core degree path
Who these are for
Fourteen guides, one per university module, all to the same standard. Read one and you know what the rest will be like. Every guide has a free sample chapter at fullmarkspress.com, and the full guides are free to read with a Kindle Unlimited membership.
Undergraduate and taught-masters students sitting the module Practitioners and career changers who need the foundations Lecturers and librarians choosing an exam-focused reading-list text Free with Kindle Unlimited, or buy the Kindle edition or paperback
Written for the exam, not the reading list
One guide per module, from first-year maths to final-year deep learning
A textbook teaches a subject. A revision guide has a narrower job: getting you to the mark scheme. These are built for the fortnight before a paper, when what you need is not another chapter of theory but the shape of a full-marks answer and the name of the trap you are about to fall into. Prerequisites are stated rather than assumed, so you know before you start whether a guide expects you to be comfortable with matrices or is going to build them for you.
Worked examples annotated the way an examiner annotates them The commonest mistakes named, so you can avoid them under pressure Diagrams that carry the explanation rather than decorate the page
About Full Marks Press
An imprint of VerityAI, publishing exam companions for university AI and machine learning modules. The guides are written by practitioners who build and evaluate these systems in production, not by generalists working from a syllabus. There are fourteen, one per module, all built to the same standard, so reading one tells you what the rest will be like. Corrections from readers go into the next edition.