The web is a visual amusement park, full of colorful images, playful gifs, and funny videos. But when that visual richness is combined with text, people with low vision or color blindness increasingly struggle with the basic function of reading and thus can’t fully enjoy what the web has to offer. In this talk, we will dive into the science of colors and how different people perceive them. We will discuss how color contrast affects text readability, and explain the challenge of estimating color contrast for text on complex background of an image or gradient. We will also explore how modern deep learning tools such as OCR and text detection can be used for text classification based on readability. In addition, I will share my personal experience of working with visual data while having a visual impairment.
Asya Frumkin is an algorithm developer, specializing in computer vision. She is also a proud Albino and visually impaired from birth. However, her disability never stopped her and gave her the irresistible drive to make a real impact. After building models in the automotive and medical imaging industries, it was fairly natural for her to join Evinced, a startup focused on improving digital accessibility. At Evinced she is responsible for end-to-end data solutions that help people like her take an equal part in the online world.