Inspiration
Every day, people suffer from skin conditions varying from ringworm circles to skin cancer. However, many of these people may have these conditions for many days before actually noticing the irregularity. With this app, we hope to make it much simpler for people to check their skin irregularities before they are exacerbated over time. With a quick snap, this app could potentially save the lives of people all around the world.
What it does
Once the person takes a picture of the skin irregularity, it is compared against the various sample images of malignant and benign skin conditions. The resulting output provides a confidence level as to how likely the skin abnormality is to be a certain disease.
How we built it
We used a python script to download images of skin conditions from the internet. Then we trained the IBM Visual Recognition neural net using these images. Finally, we made an application with C# for users to input their images in order to diagnose their skin condition.
Challenges we ran into
- Learning how to use IBM's API.
- Figuring out how to use the script to download images.
- Developing the code for sending http requests and parsing the response.
Accomplishments that we're proud of
- Good accuracy in predicting the skin condition for the skin conditions that we trained for.
- Learning various new technologies like using IBM Bluemix.
What we learned
- How to use IBM's API.
- How neural networks work.
- How to send http requests and receive responses.
What's next for SnapDoc
- Implement a mobile version of the application.
- Deploy it as a web application.
Built With
- bluemix
- c#
- ibm-watson
- machine-learning
- python
- visual-recognition
- visual-studio
- windows-form
Log in or sign up for Devpost to join the conversation.