Animal Identification with AI

Students will explore how conservationists use artificial intelligence as a tool for animal identification.

Background Information

Additional Resources 

• Learn more about how GCF uses Giraffe Spotter from WildMe 
• Take a look at Giraffe Spotter 
• Examine another way GCF tracks giraffe with Twiga Tracker
• Did you know that there are four different species of giraffe? Learn more with this Giraffe Pattern Poster
• Giraffes are one of the most endangered large mammals on the planet. Learn how to raise awareness on World Giraffe Day
• Play this Giraffe Spotter Game online 

Materials 

• The Tech for Global Good videos:
  • WildMe “The Problem” video (2:22 min)
  • WildMe “The Solution” video (2:52 min)
• Device to present the video to the class
• Bounding Boxes handout (1 per team)
• Pen or marker (1 per student)
• Giraffe Photos
  • Option 1: physical Sorting cards (print 1 set per team) 
  • Option 2: Padlet version (1 device and link per team)
• Giraffe Answer Key 

Preparation

1. Print the Bounding Boxes Handout (1 per student).
2. If you are using the physical version of the giraffe photos, print the Sorting Cards and cut enough so there is one set per team.
   • Plan to divide students into teams of 3-4. 
3. If using the Padlet version of the giraffe photos set-up the Padlet to share with student teams for collaboration and editing. 
   • Tip: Team size can be 2-3 if preferred for this version.  
4. Watch and review all of the resources to become familiar with the content.
5. Try the activity yourself, with other educators or kids you know. This will give you practice with the materials and tools to be able to anticipate student questions.

Lesson Directions

Frame the Challenge

Activate Prior Knowledge (10 min) 

1. Start by exploring what learners already know about the current decline of wildlife populations on Earth.
   • Guiding Questions could include:
     • What are some examples of animal species that are experiencing population decline?
     • How do scientists decide when a species has become endangered?
2. Explain that today they are going to be helping conservationists, or people who work to protect the environment and wildlife. They will be exploring how conservationists use data to do their work and some of the challenges they face.
3. Ask learners how they think conservationists gather data on the animals they study. Some of the responses may include:
   • Observing wildlife populations.
   • Setting up cameras.
   • Tagging animals and releasing them.
4. Have learners consider what all these types of data gathering have in common.
   • If they do not identify it, let them know that they all require the assistance of technology to be effective at gathering data.
5. Let learners know that conservationists don’t just observe large groups of species - it is also essential for them to track individual animals within the groups to learn if the total number of animals in a group is growing or shrinking.
6. Show learners the Tech for Global Good Wild Me videos: The Problem (2:22 min) and The Solution (2:52 min).

7. Follow up by asking some Guiding Questions.

• How does tracking individual animals help conservationists learn what is happening within a larger population?
• How are conservationists able to identify the individual animals they are tracking?
• Do you think there ways conservationists could study wildlife without harming the animals they are tracking?

8. Introduce artificial intelligence (AI) as a tool that many conservationists are currently using to identify, collect and analyze both individual animal’s data and large sets of data on the animal populations.

• Note that in this activity, they are going to be exploring machine learning, a branch of AI where the goal is to create a program that improves as it processes more data.

Introduce Part 1 (5 min) 

1. Introduce the scenario above.
2. Let learners know that they will be exploring how this machine learning program works by acting as the AI program for a selection of images.
3. They will first learn to identify each giraffe the same way the AI program did in the Solution video by drawing bounding boxes around any features that could identify the giraffes.
   • Define bounding boxes as a technique where a rectangle is drawn over an object that serves as a guideline for object recognition.
   • Since they are acting as the AI program, they will need to think like a computer.
   • They will be using a computational thinking skill called pattern recognition to determine where they draw the bounding boxes.
     • Ex: They could look for distinguishing spots on each giraffe.
       
   • Note that every giraffe can be identified by their individual spot pattern – just like a human fingerprint.

Introduce Part 2 (5 min) 

1. Introduce the new scenario above.
2. Let learners know they (the AI program) will now use their training from the bounding boxes to sort through a sample of image data.
3. Show learners either the physical Sorting Cards or the Padlet version of the giraffe images.
   • Let them know that they will be working in teams to sort the cards into groups.
   • They will be continuing to use the computational thinking skill pattern recognition to determine how they want to organize the cards.
4. Emphasize that since they are the AI program, they will need to decide how the cards are sorted. It’s up to them to decide the best way to group the cards to discover if/how the population has changed in the past year.

Image Sorting Activity (10 min) 

1. Pass out physical Giraffe Sorting Cards or have learners open the Padlet version of the image cards.
2. Walk around the room and observe the teams as they decide how to sort the cards.
3. Ask open-ended questions to encourage learners to consider how an AI program might organize the pictures.
   • What patterns might the AI be using to sort the images?
   • What identifying qualities did you notice in the baseline giraffes?
4. Bring the class back together when time is up.

Share-out (5 min) 

1. Ask for a few teams to volunteer how they chose to sort their cards. Have them to share:
   • The categories and how many cards are in each.
   • If there were any outliers, or cards that didn’t fit into any of the categories.
   • Whether their sorting suggests that the giraffe population increased, decreased, or stayed the same.
2. After a few teams have shared, ask learners to compare the results.
   • What were common groupings across teams? What was different?
   • Were conclusions consistent on how the giraffe population has changed? Did some AIs lead to different results?
   • Which data was not useful? Is there other image data you think could have helped you as the AI?

Debrief (5 min) 

1. Lead a short debrief of the activity and the concepts. Help students make connections between the skills they used in this activity and that of AI systems.
2. Possible Debrief Questions include:
   • Do you think the AI could learn how to do this better than a human, based on how well you did today?
   • What are the benefits and limitations of using image data to track wildlife species?
   • How else could bounding boxes be used to train an AI?

Standards Connections

Vocabulary