This iOS application is used for identifying melanoma skin cancer without visiting a doctor. We enhanced the accuracy of melanoma detection and optimized the iOS app performance.
Here we developed computer vision algorithms for identifying the size of the image, its boundaries, also, we made the model not to pay attention to extra elements (such as hair on skin, flashes, color disks, etc.), as well as to improve quality of blurred images, so that the detected object had a sharp contour.
Also, we ported Python libraries to C++ and enhanced OpenCV library algorithms of objects recognition and image processing.
We used Python language in combination with the scikit-image library for receiving PoC. For the algorithm development we worked with scikit-image library.