For some people reading the tiny text in a medicine bottle is a challenge, or some people want to know information about a medicine instantly with just waving it in front of their camera. Also, nearly half of all American adults(90 million people) have difficulty understanding and using health information, and there is a higher rate of hospitalization and use of emergency services among patients with limited health literacy. Knowing about the medicine you intake is extremely crucial and MedLook aids people in learning about their medicine, allowing them to use the correct medicine at the correct time. So we created Medlook, a simple image-recognition python based website as the future of receiving medicine information.
What it does
Simply show your medicine to a camera and submit the picture! Instantly, get information on that particular medicine, with just a click. Powered by our image-recognition model.
How we built it
MedLook is a python based web application. So, to create it we used Python 3, HTML, CSS, and Flask. We trained the medicine recognition model also on python with the help of tensorflow.
Additionally, to go along with our web application we created a website to make it easier for our users to use MedLook. We constructed our website using HTML, CSS, and Bootstrap.
Challenges we ran into
We had to make a machine learning model within a short time span, and we had a lot of problems with deploying it live online. But thankfully we ended this hackathon with a usable prototype.
What we learned
We both were not experienced in using Flask with python, so by the end we definitely developed a pretty good understanding of Flask.
What's next for MedLook
In the near future, we plan on implementing a live web-cam recognition. Additionally, in the future MedLook will be connected to a variety of databases to expand its search for information, it will have a user login/register, and it will maintain logs of all the medicines the user has uploaded files of, in case the user wants to view it again for a future reference.
Click choose file and enter a picture of medicines that are in our database. (unfortunately because of time we could only train it on a few) get some example pics here http://bit.ly/medlooksample
Click upload and you're done!