Inspiration

With the need for more advanced and accessible technology for medicinal purposes, we decided to pursue an algorithm that is able to tell if cells contain traces of malaria. With new technologies that allow you to easily sample cells, detecting cells with or without malaria can now easily be done with computer vision! We hope that contributions to helping people all around the world with our open-source code, with a simple input of a picture, this code can save lives in the future.

What it does

It takes in a .png image as a parameter to detect if that certain cell has malaria or not. Basically returns the possibility of including traces of malaria or not based on the microscopic view of the cell.

How We built it

Initially, we took the dataset "Malaria Cell Images Dataset" from [https://www.kaggle.com/iarunava/cell-images-for-detecting-malaria] and selected certain images from the folders to use. From there, we converted all the images into an array, and sorted it into training or testing, and from there, either images or labels(labels being 0 is infected and 1 is uninfected). After sorting this data into respective NumPy arrays, we fed the data into a neural network to output 0 for infected or a 1 for uninfected.

Challenges We ran into

We had trouble with resizing images, as the images from the dataset came in various pixel widths and lengths. Changing the width and length of pixels would cause lots of distortion, which altered the results of our entire code. Fixing the distortion was very difficult, especially finding the right resizing method with interpolation.

Accomplishments that we're proud of

We were able to truly train a machine learning model to detect malaria in cells, with the help and combination of many different libraries

What we learned

We learned about computer vision and especially learned the effectiveness of the machine learning library of TensorFlow. This project has given us intuition on future projects, using machine learning libraries such as TensorFlow 2.0, as we go on to create machine learning projects from scratch ourselves.

What's next for Malaria Cell Detection

We would like to try to launch an application and connect the code with a server for easier access. Our next venture is to make this project worldwide and open source for the best access to everyone.

Built With

Share this project:

Updates