Changing lives by making recommendations about problems that you face.
If you're worried that you might need braces, but dread the thought of going to the dentist. Or if your face or your child's, seems to be developing some problem that might need medical attention that isn't immediately available.
What could be invaluable is that you have an approximate diagnosis of what's wrong, and that's where our product comes in. From within a robust web application, you can take a photograph (or upload one) using the webcam, and thanks to our deep learning algorithms, receive a diagnosis that can help you pinpoint the right problem you might be facing. This can help you go to the right practitioner and can save you some considerable amount of confusion.
To build a web application that could accept photographs through the webcam and predict the following:
- Crooked teeth that might require dental care
- Cleft Lips
- Facial Palsy
- Hairy Nevus
The last three may require care from a surgeon or dermatologist, and especially in areas where access may be limited, it may be a good idea to have a primary diagnosis.
We used tensorflow for image classification using the inception v3 architecture, that allowed us to perform Deep Learning on datasets that we got from using a selenium web driver based crawler that extracted images from Google Images.
We also embedded a webcam window into our app that users can use to take photographs to be categorized.
- Tensorflow
- flask for the web app
- Selenium for web crawling
- bash for some automation tasks
- Javascript (AJAX) for webcam connections