I was inspired by the Space Potato Theme and the Potato 101 and Intro to AI Workshops to create this Machine Learning Model, which can tell the difference between healthy, edible poisonous potatoes from green, poisonous potatoes.
What it does
Essentially, my Machine Learning Model asks the user for the image address of the potato image. After receiving this link, it will then scan the potato image and compare the patterns on this potato image from the patterns on the trained images. After finding patterns on the trained images, it will then predict whether the potato is healthy or poisonous. After predicting it, it will then tell the user whether the potato is healthy or poisonous so the user will know whether it can be consumed.
How we built it
I used both the Clarifai and Cognimate software to develop my Machine Learning Model. For the Machine Learning Model to be able to predict whether a potato is healthy or poisonous, it must be trained. So, to train the model, I searched up various pictures of both healthy and poisonous potatoes so the model knows the patterns of each type of potato. It is important to search up different potato shapes, sizes, colors, so the model knows the differences and similarities between each type of potato. The model predicts whether a potato is healthy or poisonous by using the patterns from the trained images and compares them to the potato image the user gives. After training the model, I then had to test it to see whether there were any issues and to see if it was properly able to predict the potato was healthy or poisonous. After creating the Machine Learning Model, I then coded the machine learning model to do several actions. For example, I coded to first ask the user of a link to their potato image. Then, the Machine Learning Model scans the potato image the user gave, and then using the patterns from the trained images, the model then predicts whether the potato was healthy or poisonous. Then, it tells the user whether it was healthy or poisonous so the user will know whether it can be consumed.
Challenges we ran into
One major challenge that I ran into was that the Machine Learning Model kept giving me an error saying that the classifier failed. This would happen after I put all the trained images in the model and would click the model to train with them. I created new applications to see whether it was a problem with that application, but it wasn't a problem. I thought the model wasn't able to train with the pictures I searched up, so I searched up a whole new set of images for the model to train with it, however, that wasn't the issue. Then, I realized that the model may only accept a certain type of image such as jpg so I converted all the images to png. After converting all the images to png, it finally worked!
Accomplishments that we're proud of
I am very proud that I was able to fix this issue by myself and continued to persevere even though I kept receiving that error. Also, I am proud that I was able to incorporate the ideas I learned from today's workshops into my project even though I've never worked with a machine learning model before until today.
What we learned
From this project, I learned how to use different blocks to tell the Machine Learning Model to do various actions. From the workshops, I learned how to create a Machine Learning Model and use it in real life!
What's next for Potato Health Hack!
I want to train my Machine Learning Model with more images so I will be able to increase the consistency/accuracy rate of the model when identifying a potato image. I also want to add more actions to the code after it predicts whether the potato is healthy or not healthy.