Spot.

home screen
about Spot. screen
contact information screen
prediction tool
upload image screen
result page
figma
group photo

Inspiration

We were looking for problems in the healthcare community that might affect a wide majority of people by brainstorming ideas. Out of all our ideas; we were drawn to the ones that were about using photo recognition to discover early signs of illness. We juggled ideas of ADHD, Alzheimer's, Dementia; until we landed on Skin Cancer. We all felt a direct connection with this project idea as we have people in our lives impacted by or family history of skin cancer. There was an immediate reaction we all felt when the idea was brought up; which has driven our passion for developing Spot..

What it does

Spot. uses ML training on this dataset and this dataset of skin cancer related photos to detect potential signs of skin cancer via user submitted pictures of their skin. The AI does not give a medical diagnosis; it gives an estimated screening of your skin to refer you to a doctor or not.

How we built it

We built it using HTML and CSS for our front end; using Figma to produce our design. For our backend; we used Python with Flask and TensorFlow to handle our user inputted photo and our AI model.

Challenges we ran into

Early on our development we ran into issues related to using the Intel software and hardware to run our code. Setting up the environment proved to be our biggest challenge as some of our laptops just wouldn't comply with the software. This was just a minor issue as we were eventually able to sign in and use the Intel tools to test our project. Merging all of the branches together was also an issue, but we were able to resolve any conflicts. Considering the dataset that we trained our model on; we realized that the photos contained within primarily featured lighter skin, with few samples from people of color and other races. This was remedied with more data, but we'll get into that a little later in a section below.

Accomplishments that we are proud of

As this was some of our teammates' first hackathon, we're really proud of the amount of learning that we have achieved! We have been working with tools and languages that we haven't used as much as others; so teaching and learning from each other has proven to be the greatest experience of the entire event.

What we learned

To elaborate on the comment from the last section; we taught and learned several different frameworks and languages from each other. This includes general to extensive knowledge on: Python, CSS, HTML, JavaScript, Flask, and TensorFlow. This was also the first time any of us had worked with a remote development PC like those provided to us by the Intel team! Testing our project on our own PC's and those of the Intel team proved to us the sheer difference that the Intel AI PC's bring to the table in terms of computing power and speed.

What's next for Spot.

Depending on how the judgement of our product ends up; we hope to have a viable and launchable free product open for users to determine early signs of skin cancer without having to resort to competitors that offer solutions but at the cost of hundreds of dollars that aren't even ready to purchase. In order to eliminate any potential biases or misdiagnoses we want to add more relevant photos regarding examples from people of color and other races to our dataset, much like this website describes. Considering the implementation of this product to a wider audience; for rural individuals we want to connect with a potential "middle-man"/community helper to facilitate detections with our tool. For people in rural communities a doctor might be as many as miles away from their location; so having this tool to help identify issues before they make the trek would be invaluable.