I focused on the issue of unhealthy eating habits that contribute to large health risks, such as obesity and heart disease. I have been working on how an AI-powered robot can identify healthy foods as opposed to unhealthy foods simply by looking at an image. The intent is to provide a faster and easier method for deciding the healthiness of foods.
I completed this project largely due to the fact that many individuals eat unhealthy foods without a second thought , or even worse, often do not realize that they are consuming unhealthy foods. I also wanted to see if artificial intelligence could identify foods as quickly as a photograph to determine whether they are healthy or unhealthy. With that in mind, my goal was to create a relatively simple tool that would prompt users to eat foods that are healthier, which could in turn help with obesity, heart disease, and other negative side effects of unhealthy eating habits.
To make my project, I used Teachable Machine to train an AI model with pictures of healthy and unhealthy foods. I collected and uploaded images into two categories and trained the model to recognize them. Then, I tested the model by showing it new food images to see if it could correctly identify them. I improved the accuracy by adding more examples and retraining the model.
After evaluating my project, I found that the ai models have correctly identified healthy and unhealthy food most of the time. The more images I trained with, the better the model performed in classifying a diverse variety of food. It was exciting to see that the model was able to learn better with more training data. My designs in the app uses visual patterns to predict whether a food was healthy or not, this makes it easy for the user to quickly compare food healthiness.
My findings are important because they demonstrate that AI can help people make healthier food choices without complicated processes and in a timely manner. I learned that if provided with adequate training data, a robot can be trained to recognize a multitude of patterns in food pictures. This means we can use technology that will assist people in better dietary habits and suggests that these small tools can make a significant impact on health challenges such as obesity.
Then I could develop my project better staying with the theme of food images to help the robot learn more. I would then create i an app that would allow people take a photo of their food and get immediate feedback. If I did this again, I would typically have tested the robot using multiple real-life pictures to make it much more realistic. I could also have teaching it to suggest healthier alternates for unhealthy foods.
My STEAM teacher, Jason Herron introduced me to the trainable robot. Ms. Milagros helped me with my demonstrational video by being an actress.