Available resources (technology access, time, materials)
Skills being developed (presentation, coding, writing, design)
Audience (classmates, family, community, online)
Selection Process:
Review all project options
Identify 2-3 that interest you
Review requirements and time commitment
Discuss choice with teacher
Complete project proposal template
Get approval before starting
PROJECT OPTION 1: Video Explanation for Younger Students
Project Description
Create a 2-3 minute video explaining how AI sees images, designed for students 2-3 grade levels younger than you. Make the content engaging, accurate, and age-appropriate.
Learning Objectives
Synthesize complex information into simple explanations
Use analogies and examples appropriate for younger audiences
Demonstrate mastery through teaching
Requirements
Content Must Include:
✓ What a pixel is (with visual examples)
✓ How AI processes images differently than humans
✓ At least one real-world example they'd recognize
✓ Accurate information (no misconceptions)
✓ Age-appropriate vocabulary and pacing
Technical Requirements:
✓ 2-3 minutes in length
✓ Clear audio (use microphone or quiet space)
✓ Visual elements (drawings, photos, screen capture, props)
✓ Title screen with topic and your name
✓ Credits acknowledging any resources used
Presentation Requirements:
✓ Speak clearly at appropriate pace for audience
✓ Use engaging tone (enthusiastic, not monotone)
✓ Look at camera when speaking
✓ Use simple language and define any technical terms
Resources Needed
Device with camera and microphone
Video editing software (iMovie, WeVideo, Windows Video Editor, Canva)
Design a detailed proposal for a new AI image classification system that would solve a real problem in your school or community. Include technical specifications, benefits, concerns, and implementation plan.
Learning Objectives
Apply computer vision concepts to authentic problem-solving
Analyze feasibility and implications of AI solutions
Demonstrate systems thinking
Requirements
Proposal Must Include:
1. Problem Statement (1 paragraph)
What problem exists?
Who is affected?
Why does it matter?
2. Proposed Solution (2-3 paragraphs)
What would your AI system do?
What would it classify or identify?
How would it work technically?
What training data would be needed?
3. Technical Specifications
Required hardware (cameras, sensors, computers)
Software/tools needed
Estimated accuracy required
Processing speed requirements
Data storage needs
4. Training Data Plan
How many images needed?
Categories/classes to recognize
Where/how to collect images
How to ensure diversity and quality
5. Benefits Analysis
Who benefits and how?
What problems does it solve?
Time/money/resources saved?
Improved safety, efficiency, or access?
6. Concerns and Limitations
Privacy issues
Potential for bias
Cost considerations
What might go wrong?
Error tolerance
7. Implementation Plan
Step-by-step timeline
Resources needed
People/roles required
Testing plan
Evaluation metrics
8. Visual Component
Diagram showing how system works
Example use cases with images
User interface mockup
Format Options
Written report (4-6 pages)
Slide presentation (15-20 slides)
Video pitch (5-7 minutes)
Website or digital portfolio
Example Project Ideas
Wildlife identification system for school nature trail
Plant disease detector for school garden
Recyclable material sorter for cafeteria
Library book organizer using cover recognition
Lost and found item identifier
Sports equipment inventory tracker
Art room supply organizer
Accessibility tool for visually impaired students
Assessment Rubric
Criteria
Points
Problem clearly defined and meaningful
15
Solution is feasible and well-explained
20
Technical details accurate and thorough
20
Benefits analysis thoughtful and specific
15
Concerns/limitations addressed honestly
15
Visual components clear and professional
10
Overall organization and presentation
5
Total
100
PROJECT OPTION 3: Comparative AI Analysis
Project Description
Test and compare multiple image recognition AI systems using the same set of test images. Document accuracy, strengths, weaknesses, and provide recommendations about which system works best for different purposes.
Learning Objectives
Develop evaluation and comparison skills
Understand that not all AI systems are equal
Practice scientific methodology
Requirements
Systems to Test (choose 3-4):
Google Lens (identify objects, landmarks, text)
Google Teachable Machine (custom-trained)
Microsoft Seeing AI (describe scenes)
Camera apps with object detection
Social media filters with face tracking
Other accessible image recognition tools
Test Images (prepare 15-20):
Mix of:
Clear, simple objects (should be easy)
Ambiguous images (challenging)
Partial objects (incomplete view)
Poor lighting (difficult conditions)
Unusual angles (tests flexibility)
Similar objects (tests discrimination)
Data to Collect:
For each system and each image, record:
Classification/label given
Confidence score (if provided)
Time taken
Correct or incorrect?
Notable behaviors or errors
Analysis Requirements:
✓ Calculate accuracy rate for each system
✓ Identify what each system does well
✓ Identify common errors or weaknesses
✓ Compare speed and usability
✓ Note any bias or limitations observed
Deliverable:
Create comparison report with:
Executive summary (1 page)
Methodology explanation
Data tables showing results
Charts/graphs visualizing findings
Analysis and conclusions
Recommendations for use cases
Sample Data Collection Sheet
Image
Google Lens
Teachable Machine
Seeing AI
Correct Answer
Dog photo
"Golden Retriever" (95%)
"Dog" (87%)
"A golden dog sitting"
✓
Muffin
"Food" (73%)
[Not trained]
"A muffin"
✓
Partial view
"Unknown"
"Uncertain" (42%)
"Object"
Partial
Assessment Rubric
Criteria
Points
Appropriate systems selected and tested
15
Sufficient test images with variety
15
Data accurately collected and organized
20
Analysis thorough and insightful
25
Visual presentation of data (charts/graphs)
15
Recommendations clear and justified
10
Total
100
PROJECT OPTION 4: Art That Tricks AI
Project Description
Create artwork, crafts, or photographs specifically designed to confuse or trick image recognition AI. Document the AI's responses and explain why your approach worked (or didn't).
Ambiguous combinations (objects that look like other objects)
Extreme close-ups (too close to recognize)
Pattern interference (backgrounds that disrupt recognition)
Color Manipulation:
Remove expected colors (black and white photo of usually colorful object)
Add unexpected colors (green stop sign)
High contrast (loses mid-tones and detail)
Neon/unnatural colors
Context Confusion:
Objects in weird places (shoe on a plate)
Unexpected scales (giant pencil, tiny car)
Upside down or rotated objects
Multiple objects overlapping
Abstract Art:
Looks like something but isn't (pareidolia - seeing faces in random patterns)
Combines features from multiple objects
Fragmented or deconstructed objects
Requirements
Create:
✓ 5-10 original artworks/photos designed to trick AI
✓ Test each with at least one AI system (Teachable Machine, Google Lens)
✓ Document AI's response for each piece
Document:
For each artwork, record:
What you created and why
What AI said it was
Confidence score (if given)
Your explanation of why AI got confused
What this teaches about AI limitations
Present:
Create gallery presentation:
Display artwork
Show AI's response
Explain the trick
Connect to lesson concepts
Format Options
Physical gallery walk with printed photos/artwork
Digital portfolio or website
Video tour of your art with explanations
Slide presentation with images
Assessment Rubric
Criteria
Points
5-10 original artworks created
20
Clear strategy for tricking AI
20
AI responses documented thoroughly
20
Explanations demonstrate understanding
25
Creativity and artistic quality
10
Presentation clear and engaging
5
Total
100
PROJECT OPTION 5: Community Problem-Solving with AI
Project Description
Identify a real problem in your school or community that could be helped by image recognition technology. Research the problem, design a solution, create a prototype or proof-of-concept, and present your findings to a real audience.
This is an Extensive Project
Time Required: 6-10 hours over 2-4 weeks
Collaboration: Can be done individually or in teams of 2-3
Audience: Present to teacher, class, principal, or community members
Project Phases
Phase 1: Problem Identification (1-2 hours)
Interview people to find pain points
Research existing solutions
Choose specific problem to address
Get teacher approval
Phase 2: Solution Design (2-3 hours)
Plan how image recognition could help
Research technical feasibility
Consider costs, ethics, limitations
Create detailed design document
Phase 3: Prototype Development (2-3 hours)
Build proof-of-concept using Teachable Machine
Collect training images
Test and refine
Document results
Phase 4: Presentation (1-2 hours)
Create presentation materials
Practice delivery
Present to authentic audience
Gather feedback
Requirements
Problem must be:
Real (actually exists in your school/community)
Significant (matters to people)
Appropriate for image recognition solution
Feasible for student to address
Solution must include:
Technical specifications
Training data plan
Working prototype or detailed mockup
Implementation timeline
Cost/resource analysis
Consideration of ethical concerns
Presentation must include:
Problem explanation with evidence
Solution demonstration
Results from testing prototype
Discussion of benefits and limitations
Next steps or recommendations
Q&A with audience
Example Problems
Identifying native vs. invasive plants in school garden
Sorting recycling in cafeteria more accurately
Helping PE teacher track equipment
Organizing lost and found items
Identifying accessibility barriers in school building
Monitoring wildlife in school nature area
Assessment Rubric
Criteria
Points
Problem is real, significant, and well-researched
15
Solution is feasible and well-designed
20
Prototype demonstrates key concepts
20
Ethical/practical concerns addressed
15
Presentation clear and professional
15
Responds well to questions
10
Shows initiative and effort
5
Total
100
PROJECT OPTION 6: AI Ethics Research and Presentation
Project Description
Research a specific ethical issue related to image recognition AI (facial recognition, bias, privacy, deepfakes, surveillance, etc.), analyze multiple perspectives, and present findings with your own informed opinion.
Learning Objectives
Develop research and analysis skills
Consider multiple viewpoints on complex issues
Form and defend evidence-based opinions
Understand societal implications of technology
Topic Options
Facial Recognition in Schools:
Should schools use facial recognition for attendance, security, or lunch payment?
Bias in AI Systems:
How does bias in training data lead to unfair outcomes? What are solutions?
Privacy vs. Security:
How should society balance surveillance capabilities with privacy rights?
Deepfakes:
What are the risks of AI-generated fake images/videos? How can we address them?
AI in Hiring:
Should companies use image/video analysis in hiring decisions?
Social Media Content Moderation:
How should AI identify and handle inappropriate images?
Medical Imaging AI:
Should AI make medical diagnoses, and what are the implications?
Wildlife Conservation:
How does AI image recognition help or harm conservation efforts?
PROJECT OPTION 7: Create Your Own Training Dataset
Project Description
Create a high-quality, well-organized training dataset for a specific image classification task. Document your process, including how you ensured diversity and quality, then use it to train an AI model and evaluate results.
Learning Objectives
Understand importance of training data quality
Develop data collection and organization skills
Experience the complete AI training pipeline
Requirements
Choose Classification Task:
Select something meaningful and feasible:
Good: School supplies (pencil, eraser, ruler, scissors, etc.)
Good: Local birds or plants found in your area
Good: Types of weather conditions
Too broad: All animals (too many categories)
Too simple: One object in different lighting (not enough variety)
Collect Images (minimum 50 per category, 3-5 categories):
✓ Take your own photos (preferred) or use copyright-free sources
✓ Include variety: different angles, lighting, backgrounds, distances
✓ Ensure high quality: in focus, clearly shows subject
✓ Organize in folders by category
✓ Name files consistently (bird-cardinal-01.jpg)
Document Dataset:
Create "README" file including:
Purpose of dataset
Categories and number of images each
How images were collected
Diversity considerations
Known limitations
How others should use it
Train and Test Model:
Use dataset to train model in Teachable Machine
Test with new images not in training set
Document accuracy and errors
Analyze what worked well and what didn't
Share (Optional):
If age-appropriate and with permission:
Upload to educational dataset repository
Share with classmates for their projects
Contribute to citizen science project
Quality Checklist
For Each Image:
Clear subject (not blurry)
Well-lit (can see details)
Appropriate background (not distracting)
Copyright-free (you took it or have permission)
Correctly labeled and filed
For Overall Dataset:
Balanced categories (similar numbers of images each)
Diverse examples in each category
No duplicates
Consistent quality standards
Well-organized file structure
Documentation complete
Assessment Rubric
Criteria
Points
Appropriate task selection
10
Sufficient images collected (50+ per category)
20
Diversity and quality of images
25
Organization and file management
15
Documentation thorough and clear
15
Training results and analysis
15
Total
100
PROJECT OPTION 8: Computer Vision Career Research
Project Description
Research careers that involve computer vision or image recognition, interview someone in the field (in person, video call, or email), and create a career profile presentation sharing what you learned.
Learning Objectives
Explore potential career paths
Understand real-world applications of concepts
Practice professional communication
Connect learning to future opportunities
Careers to Research (choose 1-2)
Technology Careers:
Computer Vision Engineer
Machine Learning Engineer
AI Research Scientist
Software Developer (specializing in image processing)
Data Scientist
Robotics Engineer
Applied Careers:
Medical Imaging Specialist (Radiologist, MRI Technician)
Autonomous Vehicle Engineer
Quality Control Inspector (using AI)
Wildlife Biologist (using camera traps)
Digital Forensics Analyst
Agricultural Technology Specialist
Security Systems Specialist
Creative/Design Careers:
Visual Effects Artist (using AI tools)
AR/VR Developer
Game Developer (implementing computer vision)
Computational Photographer
Requirements
Research Component:
✓ Job description and typical responsibilities
✓ Required education and skills
✓ Typical salary range and job outlook
✓ How computer vision is used daily
✓ Challenges and rewards of the work
✓ Career pathway (how to get there from where you are)
Interview Component:
✓ Find and contact professional in field
✓ Prepare 10-15 thoughtful questions
✓ Conduct interview (15-30 minutes)
✓ Take thorough notes or record (with permission)
✓ Send thank you note after
Presentation Requirements:
✓ Overview of career
✓ Day-in-the-life description
✓ Interesting facts from interview
✓ Skills needed and how to develop them
✓ Why you would/wouldn't pursue this career
✓ Resources for learning more
Interview Questions to Ask
About the Work:
What does a typical day look like for you?
What projects are you currently working on?
How do you use computer vision/image recognition in your work?
What's the most interesting problem you've solved?
About the Field:
How has this field changed since you started?
Where do you see the field going in the next 5-10 years?
What breakthroughs are most exciting to you?
About Skills and Education:
What education and training did you need?
What skills are most important in your work?
What do you wish you had learned in school?
How can middle/high school students prepare?
About Career:
How did you become interested in this field?
What do you love most about your job?
What's most challenging?
What advice would you give to someone interested in this career?
Finding Professionals to Interview
Where to Look:
LinkedIn (search for job title in your area)
Professional associations (IEEE, ACM, etc.)
University professors (computer science, engineering)
Local tech companies
Online communities (Reddit, Discord in relevant fields)
Family friends or connections
Teacher connections
How to Reach Out:
Email professional introduction
Explain your project
Request 15-30 minute interview
Offer flexible scheduling
Be professional and courteous
Assessment Rubric
Criteria
Points
Research thorough and accurate
20
Interview conducted professionally
25
Interview questions thoughtful and relevant
15
Presentation engaging and informative
20
Personal reflection and connection
10
Professional communication throughout
10
Total
100
GENERAL PROJECT TEMPLATES
Project Proposal Template
Name:Date:
Project Option Selected:
Project Title:
Why I chose this project:
What I hope to learn:
Resources I'll need:
Timeline:
Start date:
Major milestones:
Completion date:
How I'll present my work:
Written report Presentation Video Website Other: _______
Teacher approval: Date:
Project Planning Sheet
Week 1 Goals:
Week 2 Goals:
Week 3 Goals:
Challenges I'm facing:
Help I need:
Project Self-Reflection
What I learned about computer vision:
What I learned about myself:
What I'm most proud of:
What I would do differently:
How this connects to my interests or future:
Evolve AI Institute • Lesson 7: How AI Sees Images
Extension Project Guidelines and Templates
Choose a project that excites you and deepens your understanding. These projects let you explore computer vision concepts in creative, meaningful ways!