Tobias Beidler-Shenk, Steven Johns, Jeff Zheng, Kavin Sankar
Age related visual impairment is a major health problem among the elderly. With advancing age, the normal function of eye tissues decreases and there is an increased incidence of some form of visual impairment. The most common causes of age related visual impairment in the elderly are cataracts, which affect 1 in 5 adults over the age of 65, and presbyopia which affects over 85 percent of adults over 40. With statistics like this, it seems almost inevitable that someone could take the mistakenly take the wrong pill. For Senior Citizens, taking the wrong pill could be life-threatening. Our project aims to eliminate that concern for every senior citizen. With our Pill Identifier, anyone can get the name and information about the pills they take by simply taking a picture of it.
The Pill Identifier classifies pills using a machine learning model. Our model was trained on a relatively small dataset from the CDC that we extrapolated using image rotation, color, blurring, contrasting, among others. As of now, the algorithm only supports 8 different classes of pills, but given a larger dataset and more time, we would generalize to many more classes.
On the front end, we used html, css, and javascript to make a simplistic UI, as our target demographic is less familiar with technology. The application is optimized for mobile, making image upload much simpler. We made the decision to use large fonts and high-contrast colors to make sure that our visually impaired users can effortlessly learn about their pills.