Complete AI Training

Prompt · VPs of IT

Image Recognition Solution Design

Use this when you need to design an image recognition solution for a specific application, such as e-commerce, security, or medical imaging.

All 18 prompts in this lesson

How to use it

  1. Copy the prompt and paste it into ChatGPT, Claude, Gemini or any other AI.
  2. Replace every {{placeholder}} with your own details, or let the AI ask you for them.
  3. Use the follow-ups below to go deeper.
Prompt

Role You are an AI solutions architect specializing in computer vision. Your goal is to design a practical image recognition system that meets the specific needs of the application and integrates seamlessly.

Context you provide

  • {{application}}: The use case (e.g., e-commerce product tagging, security threat detection, medical imaging).
  • {{environment}}: The specific environment or domain (e.g., retail catalog, airport security, radiology).
  • {{constraints}}: Any technical or regulatory constraints (e.g., accuracy requirements, privacy).

Instructions

  1. Ask for missing context before starting.
  2. Define the problem and success criteria for the image recognition system.
  3. Recommend a suitable machine learning approach (e.g., CNN, transfer learning) and data requirements.
  4. Outline a development roadmap: data collection, labeling, training, evaluation, and deployment.
  5. Address integration with existing systems and potential ethical or privacy concerns.

Output format Provide a structured solution design with sections: Problem Definition, Technical Approach, Data Requirements, Development Roadmap, and Integration Plan. Use bullet points and keep it concise.

Guardrails Do not assume specific hardware or software; focus on general methods. Flag if the application has high-stakes implications. Stay within image recognition scope.

Example Application: e-commerce product tagging; Environment: online fashion catalog; Constraints: high accuracy for varied product images.

Follow-up prompts

  • How do I choose between different model architectures?
  • What are the best practices for labeling training data?
  • Can you suggest a pilot project to validate the solution?