I remember the mayhem that the COVID-19 outbreak caused when it first broke out. Every day on the news I heard that there was a shortage of medical professionals to diagnose and treat people who had contracted the coronavirus. This year, the monkey pox outbreak has caused a similar level of public outcry. Given that the symptoms of monkey pox are visual though, (skin lesions) I got the idea to create a binary image classifier that can distinguish between skin lesions originating from monkeypox versus other causes. I hope that this ML model can be utilized so that people will be able to diagnose themselves using technology without overwhelming the already strained medical systems.

What it does

My software is a binary image classifier model that was trained to distinguish between skin lesions originating from monkey pox versus other causes.

How we built it

The binary image classifier was created using a Convolutional Neural Network (CNN) model through the Tensorflow and Keras libraries. The images in the dataset from kaggle were first converted into a NumPy array through the scikit-image library. Keras was then able to create a neural network taking the NumPy array of the dataset images as its input. The model was trained and it's efficiency was tested using the accuracy function.

Challenges we ran into

The biggest struggle I had while working on this project was Tensorflow. This was the first time I had worked with Tensorflow so I didn't know how to use it that well. Additionally, Tensorflow didn't seem to work on so I had to do the second part of my project on Google Colab. With a lot of help from stack overflow discussions though, I was able to complete my project at the end!

Accomplishments that we're proud of

I'm really proud of learning how to use Tensorflow and the technical aspect of this project, but I'm most proud of the idea that I came up with. This is my first hackathon submission where I genuinely feel like my solution can be developed and implemented to have a real impact on the world and solve pressing issues.

What we learned

I learned how to use Tensorflow and Keras libraries. I also fine-tuned my skills with scikit-image and I was also able to get a firm understanding of the python os library which helped me with file management throughout.

What's next for Monkey Pox Lesion Detector

I hope to find a bigger dataset so that my ML model accuracy can surpass 90% (a couple thousand images will be needed) And then, I intend on integrating this ML model into a flask web application where users can go and input an image of their lesion and get a fairly accurate prediction of whether they have contracted Monkey Pox or not.

Built With

Share this project: