Library . Public Results TEACHERS: Machine Learning and Image Recognition for Weather Forecasting

The Problem

Climate Challenges is affecting various countries in Europe. These include more frequent and intense heatwaves, droughts, and wildfires, which are exacerbated by climate change. The Mediterranean region, where Greece is located, is particularly vulnerable to these effects. The summer of 2021 saw catastrophic wildfires that ravaged large parts of the country, destroying forests, homes, and livelihoods.

Educational Scenario of Scientific Action

The educational scenario of scientific action “Machine Learning and Image Recognition in the Service of Climate Change” prepares students to create an application in which, with the help of a virtual drone, they can record clouds formation and classification using machine learning. Certainly, the application could be extended to recognise other situations, such as fires, the melting of glaciers, etc.

On the path to achieving the goal, students learn what artificial intelligence is, how it can be used in our daily lives, engage with issues of personal data protection, ethical dilemmas that arise, etc.

Sustainable Development Goals

13. Climate Action
15. Life on Land
16. Promote peaceful and inclusive societies

Curriculum

Curriculum – Scenario Connection

The scenario can be implemented in all middle school grades within the context of the Informatics course. If taught in the 1st Grade, it should be after some lessons in Scratch have preceded.

Activity Content

1. Scenario Presentation
2. Scenario Description
3. Teacher’s Guide
4. Care – Activities
5. Know – Activities
6. Do – Activities
7. Activity sheet for discussion at home with parents
8. Machine Learning for Kids- Instructions
9. Videos
10. Scratch code for wild animals
11. Scratch code for wild animals in images
12. Scratch code for Amazon deforestation
13. Scratch code for Amazon deforestation in images

CARE-KNOW-DO activities:

There are separate activities for the units Care – Know – Do which lead us to our goals.
Follow the activities designed with four learning objectives:

1. Understanding the scientific framework (2. Care – Activity) – «Νοιάζομαι»
2. Familiarization with a machine learning application (3. Know – Activity) – «Μαθαίνω»
3. Familiarization with an existing scratch program for image recognition, with a focus on endangered animals (3. Know – Activity) – «Μαθαίνω»
4. Creating a scratch program for image recognition, focusing on identifying human intervention in forests via drone (4. Do – Activity)

Participatory and Fun Science

In this educational scenario of scientific action, students, family members, scientists participate. Students are invited to identify problems and create solutions, interact with experts, and take initiatives for action to protect the threatened environment

Sources:

• https://arxiv.org/abs/1609.01775
• https://wires.onlinelibrary.wiley.com/doi/epdf/10.1002/widm.1278
• https://waymo.com/waymo-driver/
• https://www.nature.com/articles/d41586-020-03187-3#ref-CR1
• https://exposing.ai/datasets/
• https://jivp-eurasipjournals.springeropen.com/articles/10.1186/s13640-018-0391-6
• https://www.microsoft.com/en-us/ai/ai-for-earth-imazon
• https://www.mdpi.com/2072-4292/11/13/1534/htm
• https://machinelearningforkids.co.uk/
• https://teachablemachine.withgoogle.com/
• https://code.org/oceans

Supporting Material

Κείμενα

1. The advantages and disadvantages of facial recognition
2. The usefulness of image recognition
3. Harnessing_Artificial_Intelligence_for_the_Earth_report_2018
4. The problem of bias in facial recognition

Images

• Wild animals
• Amazon