This module equips the student with an understanding of and associated skills for the development of real-world robust computer vision systems. This will also bridge the gap between domain knowledge and the successful application of that knowledge in computer vision.

Part 1: - Architectural Design Elements of all Computer Vision Systems (Basic component selection and video chain) Part 2: Camera Design - Basics of optics: how does a lens work? - Application Field of View (wide automotive, narrow industrial or medical) - Lighting and light sources (controlled or uncontrolled?) - The image sensor: How does an image sensor work, colour, Colour Filter Arrays (including Bayer pattern and hyperspectral), CMOS and CCD, electronic shutter, Rolling vs Global Shutter, SPAD, noise and sensitivity considerations, pixel resolutions, IR/hybrid spectral architectures, matching a sensor to an application - Image Signal Processing: ADC, Exposure, Gain, Demosaicking, Denoise, Edge Enhancement, Control Loop structuring Part 3: The transmission medium - Digital and analogue: PAL/NTSC, Serialisation/Deserialisation, LVDS, Bandwidths, Slew, EMC, Shielding, Bit depth, bandwidths - Transmission formats: components of an analogue signal, components of a digital signal, Interlaced/progressive/flyback, resolutions, security, encryption, - Transmission over ethernet: compression (do we need all the bits?), internet connected camera, lock-loss, lock loss handling, JPEG/MPEG, IEEE1722 multimedia handling Part 4: The central processing hub - Types of processing: image enhancement - example image filters, computer vision - example face detection, image rendering, high dynamic range (HDR) imaging - example automotive top view, augmented reality - Image processing units - GPUs, DSPs, SoCs, host control, diagnostics - Memory consideration - memory bandwidth, memory size, hierarchical memories - Latency - how fast do you need to act on image data? Part 5: The onward interface (Human Viewing and Computer Vision) - Display technology, rendering for human consumption (images + overlays), human vision system (HVS), perception - Computer Vision considerations - calibration, dropout handling, determination of KPIs for an application in the context of the vision system - Storage - for many applications, we record the image for later use and consumption - Control signals - robotic, automotive control Part 6 : Computer Vision System Design - Functional Requirements Engineering for a Computer Vision System - Hierarchical Design considerations for generation of Non-Functional Requirements and Test Cases, HLDs/LLDs, Standards and SIL considerations - Stakeholders, Planning and RASICs Part 7: Formalisms of Computer Vision System Tuning, Verification, Validation, Debugging, Releasing - Reviews, Testing and Debugging - DVP&R, FMEA, PDCA, Change Requests, Sign-off, Maintenance

On successful completion of this module, it is expected the students will be able to: - Understand the components of robust real-world production-level computer vision systems - Perform critical evaluations of trade-offs and choices in computer vision system applications - Build a basic computer vision system

On successful completion of this module, it is expected the students will be able to: - Appreciate the development considerations of a computer vision system development team - Co-operate with team members in the development of computer vision systems - Appreciate the ethical implications in the deployment of a computer vision system

On successful completion of this module, it is expected the students will be able to: - Assemble the electromechanical components of a computer vision system

The module exhibits a blended approach, with a combination of lectures, discussions, and face-to-face lab and tutorial work. The form will be 2 hours of teaching per week. Face-to-face time will be in a 2 hour lab and 1 hour tutorial format. There will also be contributions from other experts in the field as a supporting activity, with references to the state-of-the-art.