Example-BK
Title
Example
Control
Technologies
Additional Features for the future
Data Science

healthE

pronounced '\hel-thē\'

Inspiration:

Kind of similar to adblocker, but rather than blocking the content you don’t want to see, we replace unhealthy foods with healthier alternatives. The average American spends ~10 hours a day in front of a screen and the content we see can have a lasting impact on the decisions we make. We may not know it but our minds can be trained to take hints here and there from the images we see, the videos we watch, the sounds we hear, and more.

What it does:

The Chrome Extension will analyze a website to pull the images and run them through the Microsoft Vision API, a Clarifai Machine Learning Model and a Naive Bayes Classifier to classify what type of image it is. If it is classified into what we have set to be as a “junk” food, then it will be replaced with a healthier alternative.

The user is in control here. The user is able tos elect what they would like to see and not see. For example, they can choose to not see fast food images and that images that would fall into the category of Fast food would be replaced with Fruit.

How We Built healthE:

The team was able to build the Chrome Extension utilizing the following technologies:

Microsoft Vision API
Clarifai Machine Learning Model
Naive Bayes Classifier
HTML5, CSS3, and JavaScript
mLab
Amazon Web Services

Challenges we ran into:

The Microsoft Vision API provides useful and specific information about the images of food. This was useful but we needed to more information to make a good decision about what the image was. Using the information from Microsoft Vision API, we further analyzed the image with Clarifai Machine Learning Models. With the extracted concepts, we used a trained model to predict if the concepts extracted from the image, and the image, is one of Soda, Chips or Fast Food.
Speed: Initial approach was to train many images to determine what was in them based on hundreds of images. For the time we had, this approach was not feasible. That is why we went with the Microsoft Vision API + Clarifai Machine Learning Models + Naive Bayes Classifier. Accomplishments that We are Proud of: It works;

What We Learned:

Understanding how the psychology of what we see and how it influences our actions and decisions.

What's Next:

In the future the following can b be added to the extension to increase the versatility of the product to be more useful in ways that could improve one’s health, since we spend more and more time in front of the screen.

Features to add:

Control: Ability to have more detailed options on what you would like to see and not to see
- For example, we would incorporate desserts as a breakdown of snacks and the user would be able to type search what they would like to replace desserts with
Idle time: this would be a feature to remind the user that they can take a break after working in front of a screen for X amount of minutes (X is determined by the user)
Blue Light: this feature would be able to detect the background of a certain website and be able to to adjust the amount of blue light that is emitted. Similar to flux, but for specific websites. Helps the user maintain good eye health
Recommendations: the user would be able to click on a replaced image and be directed to a website where they can purchase the healthier alternative