Audience
- Developers and analysts exploring image-based applications.
- Quality, manufacturing or operations professionals with basic coding confidence.
- Technical project staff who need to evaluate a computer-vision prototype.
AI & technology
Computer Vision
Explore how AI interprets images and video, from fundamentals to practical applications.
Learning with a practical purpose.
Explore how image-based AI can support visual inspection and other practical business tasks. This module begins with the images themselves: what they show, how they were collected, and whether they represent the conditions in which a proposed solution will operate. You will progress towards a modest, testable image-analysis prototype.
The central case is a visual quality-checking workflow. You will define categories, prepare an image collection, compare model approaches and examine errors caused by lighting, backgrounds or unfamiliar objects. Sessions balance hands-on implementation with the judgement needed to explain when a model is useful and when a human inspection remains necessary.
Live onlineTeaching through Kent Business College, UK
2 hoursOne live session each teaching week
14 sessionsA semester for each module
28 live hoursPer module, plus independent study
The outline below is a 14-session online module, with one two-hour live session each teaching week. Reading weeks, breaks and additional activities follow the agreed cohort calendar. The cohort syllabus, full route, dates, fees, independent study and assessment requirements are confirmed before enrolment.
Who this module is designed for.
Starting knowledge and preparation
- Basic Python and familiarity with simple data tables.
- A laptop capable of running the agreed teaching environment; specialist hardware is not a prerequisite.
- Use images you are permitted to process and share.
What you will work towards.
- Define a visual task and distinguish classification, detection and segmentation.
- Prepare an image dataset with representative labels and sensible evaluation splits.
- Adapt an existing model for a narrowly defined inspection problem.
- Evaluate performance under changes in lighting, viewpoint and background.
- Design a reviewable prototype with appropriate uncertainty and escalation handling.
Explore the 14-session learning plan.
Explore the focus and practical output of each teaching session. Open individual sessions below, or expand the complete plan.
Session 01 Choosing a visual inspection problem
Specify the decision, users and inspection conditions before selecting a model or collecting images.
Practical output: A visual-task brief and success criteria.
Session 02 Images as numerical data
Explore pixels, channels, resolution and transformations through straightforward image-reading and display exercises.
Practical output: An annotated image-exploration notebook.
Session 03 Collecting representative examples
Plan image acquisition across conditions and identify missing cases that could undermine later evaluation.
Practical output: A collection plan and dataset inventory.
Session 04 Annotation and label consistency
Write classification labels and bounding-box guidance, then inspect inconsistencies in a small labelled sample.
Practical output: A reviewed annotation guide.
Session 05 Splits, leakage and augmentation
Separate related images appropriately and test transformations that preserve the meaning of each label.
Practical output: A justified split and augmentation strategy.
Session 06 A simple image-analysis baseline
Apply basic image features or threshold rules to understand what a transparent method can achieve.
Practical output: A baseline inspection result.
Session 07 Understanding neural vision models
Trace the role of learned visual features and compare model outputs on deliberately varied examples.
Practical output: A model-behaviour observation sheet.
Session 08 Adapting an existing classifier
Run a small transfer-learning exercise and record the training choices needed to reproduce it.
Practical output: An adapted classifier and training record.
Session 09 Finding objects in an image
Explore detection outputs, overlapping boxes and missed objects using a controlled inspection scenario.
Practical output: A detection experiment and error examples.
Session 10 Measuring performance for the task
Interpret confusion matrices and task-appropriate detection measures in relation to operational consequences.
Practical output: An evaluation report with prioritised errors.
Session 11 Robustness outside the ideal image
Test blur, shadows, camera changes and unfamiliar backgrounds to reveal weak points in the workflow.
Practical output: A robustness test matrix.
Session 12 Human review and confidence handling
Define when an image should be accepted, rejected or sent for review without treating scores as certainty.
Practical output: A review workflow and decision thresholds.
Session 13 Packaging the inspection prototype
Connect image input, prediction and explanation into a simple interface that records review outcomes.
Practical output: A working visual-inspection prototype.
Session 14 Demonstration and deployment judgement
Present results, demonstrate difficult cases and propose practical conditions for a limited operational trial.
Practical output: A final demonstration and trial proposal.
Apply the learning
Visual quality-checking prototype
Create a small application that categorises product or document images and highlights uncertain cases. The project should address one clearly bounded inspection decision. Include a comparison between normal operating images and challenging examples so that stakeholders can judge the conditions under which the prototype could be useful.
Your project evidence can include
- A labelled dataset with collection notes.
- A reproducible model and inspection interface.
- An error catalogue and operational trial plan.
Use anonymised, public or suitable synthetic material where workplace information is confidential. The issued assessment brief confirms the required evidence.
Assessment, feedback and certification.
- Practical exercises assess image preparation, annotation and interpretation of model outputs.
- The prototype is judged against declared evaluation criteria rather than an unsupported headline accuracy figure.
- A project discussion examines failure cases, review arrangements and the implications of changed camera conditions.
Teaching and the relevant course certificate are provided through Kent Business College in the UK. Completion requirements are set out in the module information; attendance alone should not be assumed to meet every assessment requirement.
Questions worth asking.
Do I need a specialist camera?
No. The teaching project can use permitted existing images or ordinary camera images. The emphasis is a sound workflow and evaluation.
Does the course teach facial recognition?
The core project focuses on visual inspection. Identifying people through face recognition is not a required classroom exercise.
Will the prototype be production-ready?
You will build an evaluated learning prototype and a trial plan. Production deployment requires further testing against the organisation’s actual operating conditions.