0

Mission: Detecting Early Signs of Melanoma with Artificial Intelligence

Hello. I am building an Artificial Intelligence that will detect Melanoma (type of skin cancer) image-based with pixels. I will use Convolutional Neural Networks and image-based Classificiation. I will use Google Colab and am thinking to code this with Python using the OpenCV library. Motto: Prevention is better than cure The goal is that someone with potentially Melanoma can take a picture of the spot, and identify if it is the disease or not with great accuracy via their smartphone. Survival rate from Melanoma drops from 99% if detected in its early stages to about 14% if detected in its latest stages. Technology like this could mean saving lives. If you could recommend any tips, tricks, anything helpful please comment. I am at the beginning stages of this project, I have no experience in Artificial Intelligence or coding. That is why I am planning what to research and focus on first. Any help is appreciated!

7th Feb 2021, 12:55 PM
Sidar Sahin
6 ответов
+ 9
You'll need a lot of training data (aka images of those with early melanoma) if you're aiming to do this via photo. That would be my main concern. Hopefully you have a starting image library, or some plan or deal to take photos of those diagnosed with it. If you don't have a large dataset, you'll run into a problem called overfitting. Which is where the AI memorizes the training data instead of learning to generalize with it. Like a high school student cramming books to pass a test rather than actually learn the material. Depending on the model size you could train it for cheap or even free.
7th Feb 2021, 1:22 PM
Ahri Fox
Ahri Fox - avatar
+ 3
Okay, well take your time and ask specific questions that you want answered. "Any advice" is a blanket statement that covers literally anything, making it open ended. Just try and make the app and when you run into a roadblock and have discovered what the problem is and not why, then ask. If somebody tells you exactly what to do and how you'll do it, then they might as well be making the project
7th Feb 2021, 1:27 PM
Slick
Slick - avatar
+ 2
I would get the basics down of whatever you want to do before undertaking an enormous task seemingly by yourself. Plus, when asking for help, it's good to show something...anything that you've worked on to show that you have put time into it. It seems now like you've had a common idea and now want others to pitch in for the hard part (actually making something.) I don't mean to come across rude, but what do you expect from a begginer programmer app?
7th Feb 2021, 1:19 PM
Slick
Slick - avatar
0
Not to ruin your dreams, but this basically already exists. A person can just go on Google images and find pictures of melanoma and compare it to what they have. Infact, a person would be able to detect similarities much faster and more accurately than a computer.
7th Feb 2021, 1:20 PM
Adam Kollgaard
Adam Kollgaard - avatar
0
Slick I understand. I am now at the phase of research and learning, and my research includes asking questions so I don’t miss a part. My question was actually not app-related since I will figure that out later if I want to make an app or not. It is more Python in relation to Artificial Intelligence related. Indeed getting the basics down is what I am doing right now. So any advice is appreciated.
7th Feb 2021, 1:24 PM
Sidar Sahin
0
Ahri Fox Right now I am learning this so I know how to make an Artificial Intelligence like this. Indeed, I will need lots of data. Right now I am mapping out what is most important to learn to make this Artificial Intelligence project a reality.
7th Feb 2021, 1:27 PM
Sidar Sahin