Chetanna Ezomike explores AI Wildfire Detection on Zombie Fires in Ottawa Regional Science Fair

On March 27–28, Chetanna Ezomike, a grade 7 student took part in the Ottawa Regional Science Fair. His project, titled When Does AI Get Confused? Testing AI Wildfire Detection on Zombie Fire Images, explored the reliability of artificial intelligence in an increasingly important real-world application: wildfire detection.

As artificial intelligence becomes more widely used to monitor wildfires across Canada, Chetanna became curious about how well these systems perform when faced with unusual situations. In particular, he was fascinated by zombie fires—underground wildfires that continue smoldering beneath the surface throughout the winter before reigniting in the spring. Because these fires often look very different from typical wildfires, he wondered whether an AI model trained only on conventional wildfire images would still be able to recognize them accurately. His goal was to investigate one of AI's greatest challenges: how well it generalizes to data it has never encountered before.

To conduct the experiment, Chetanna trained an image classification model using Google Teachable Machine to distinguish between "Wildfire" and "No Fire" images. After training, he evaluated the model on several categories of images, including ordinary wildfire photographs and zombie fire images. The model achieved approximately 97.75% accuracy on regular test images, demonstrating that it had learned to recognize common wildfire patterns very effectively. However, when zombie fire images were introduced, the model's overall accuracy dropped to approximately 89%, and the zombie fire images themselves were classified correctly only 40% of the time. These findings demonstrated that although AI systems can perform exceptionally well on familiar data, they may struggle when presented with unfamiliar scenarios that were not represented during training. The project also highlighted the importance of diverse, representative datasets when developing AI systems intended for real-world use.

Through this project, Chetanna gained experience in machine learning, computer vision, experimental design, data analysis, and scientific communication. He also strengthened his ability to design fair experiments, interpret unexpected results, and explain technical concepts clearly to judges and members of the public.

Presenting his project to judges and visitors was one of the most rewarding parts of the experience. Answering questions about his methodology and explaining the significance of his findings helped Chetanna become a more confident speaker and taught him how to communicate complex ideas clearly to people with different levels of technical knowledge.

One of the most interesting aspects of the Ottawa Regional Science Fair was seeing how many students were applying artificial intelligence to solve real-world problems. Several projects used machine learning or computer vision in areas such as healthcare, environmental monitoring, accessibility, and robotics. Although each project addressed a different challenge, they all demonstrated how AI can be used as a powerful tool to analyze data, recognize patterns, and support decision-making. Seeing these projects gave Chetanna new ideas for improving his own work and showed him how quickly artificial intelligence is becoming an important part of scientific research across many different fields.

Participating in the Ottawa Regional Science Fair was an incredibly valuable experience. It showed him that success in research is not defined solely by winning prizes, but by asking meaningful questions, learning from the results, and sharing knowledge with others. Chetanna is grateful to his mentor, Dr. Minna Allarakhia, and to his parents for encouraging and supporting him throughout the project. Their guidance helped make this opportunity possible, and the experience has inspired Chetanna to continue exploring the intersection of artificial intelligence and environmental science. Chetanna looks forward to returning to the science fair next year.