Images as Arrays and Visual Data Ethics
By the end of this module you can describe an image as the array of numbers a model actually sees, and name who is in a dataset, who consented, and who is missing before any of it is used.
Computer Vision and Multimodal AI
A responsible vision workflow with image preparation, model or search pipeline, error analysis, dataset limits, and a model card.
Loading, Resizing, Color Spaces, and File Formats
By the end of this module you can load, resize and convert images without quietly destroying information - and explain why a resize that looks fine on screen can change what a model learns.
Basic Image Processing with OpenCV
By the end of this module you can apply the everyday OpenCV operations - blur, crop, mask, threshold - and say what each one removes from the image as well as what it reveals.
Edges, Contours, Thresholding, and Segmentation Intuition
By the end of this module you can find edges, contours and regions in an image, and judge when a clean-looking segmentation is really an artefact of your threshold rather than the object.
Building Image Datasets and Labels
By the end of this module you can assemble an image dataset with labels you would defend: a written labelling rule, a check for class imbalance, and an honest note on what the set does not cover.
CNN Intuition and Transfer Learning
By the end of this module you can explain what a convolutional network learns layer by layer, and use transfer learning to get a working classifier from a small dataset without pretending you trained it from scratch.
Classification Evaluation and Error Analysis
By the end of this module you can evaluate a classifier beyond accuracy - per-class performance, confusion between specific pairs, and the error analysis that tells you which failures actually matter.
Object Detection Concepts and Bounding Boxes
By the end of this module you can explain how detection differs from classification, read a bounding-box prediction with its confidence, and describe what an IoU threshold silently decides on your behalf.
Embeddings and Image Similarity
By the end of this module you can turn images into embeddings, find similar images with them, and explain why 'similar' means visually similar rather than semantically correct.
Multimodal Prompts and Vision-Language Model Limits
By the end of this module you can prompt a vision-language model effectively and state its real limits: confident descriptions of things that are not there, text it cannot read, and detail it invents to fill a gap.
Bias, Privacy, and Model Cards for Vision
By the end of this module you can identify where bias enters a vision pipeline, decide what should never be inferred from a face or body, and write a model card that a reviewer can hold you to.
Capstone: Responsible Vision Workflow
In this capstone you will build a vision workflow end to end and publish it with its dataset limits, its per-class error analysis, and a clear statement of the decisions it must not be used to make.
First-release boundary
This is a text-first course shell. Leonardo/image creation, image QC, dependency-heavy runtime QA, external review, and learner pilot evidence remain separate later quality steps.