Computer Vision - 4 Week Discovery


This 4-Week Discovery will provide a detailed feasibility assessment of developing a Computer Vision use case and requirements to create a prototype solution.

Our approach is based on our experience of delivering computer vision models on the Microsoft Azure stack, we always recommend adopting a phased approach to limit commercial investment and clearly define the value statement. The first of these phases is the Discovery.

Discovery: The Discovery phase will ensure all parties are aligned on future use case success criteria and the best approach. We know that we can make faster and more efficient progress if we take the time to ensure that we’re working on the right challenges, with the right solution delivered by the right team on the right technology. There are 3 phases within Discovery:

  1. Understand
  • Work closely with SME team and stakeholders to understand business priorities and strategy.
  • Formulate the overall vision and desired end results for prototype
  • Identify any key risks, issues or opportunities.
  • Gather information and review current process through interviews, technology and data assessment.
  • Understand device location and specification.
  1. Design
  • Create architecture design and solution design for analytical prototype solution. This may include Computer Vision & ML using Azure Machine Learning and Azure Databricks; model inference using Azure Kubernetes; tracking using MLflow; and storage in Azure Data Lake Storage V2.
  • Solution design must take into consideration longer term integration and requirements to ensure scalability.
  • Determine success metrics for project and process for testing.
  • Agree design principles and apply across the solution design
  • Apply learnings from experience to ensure appropriate solution design
  1. Validate
  • Confirm and validate design with SME ensuring solution meets business requirements and considers success metrics.
  • Determine resource required, provide/confirm accurate timelines
  • Confirm risk management plan, detailed planning, effort and estimation.
  • Ensure design is scalable for the future and where possible the model has ability to generalise.
  • Deliver defined scope of work for the prototype on the Microsoft Azure stack

Deliverables and Outcomes

  • Feasibility assessment and Go/No-Go recommendation for additional use cases.
  • A scope of work with defined plan by sprint
  • Commercial proposal for next phase (prototype)
  • List of requirements for measuring success