CVPR Daily - Wednesday
by Daniela Massiceti Consider for a moment the number of things you use your mobile phone’s camera for every day. The picture you snap of your child on their first day at school or your new puppy. The QR code you scan as you sit down for lunch at a restaurant which is fully hands-free after the pandemic. The document you take a picture of and upload for your visa application, because you don’t have a printer at home. Without realizing it, most of us rely on having a camera in our pockets as we go about our daily lives. For 100,000s of people who are blind or low-vision, a mobile phone camera unlocks another kind of superpower - the ability to receive information about the world around them. Apps like Seeing AI, TapTapSee and BlindSquare use cutting-edge computer vision models to understand and instantaneously return information about images and videos - from localizing objects, describing scenes, detecting colors, reading signs, and transcribing handwritten and printed documents. The VizWiz Grand Challenge workshop aims to galvanize a multi-disciplinary community who are passionate about driving technical innovation in computer vision to power these assistive experiences for people who are blind or low-vision. Building on three successful previous instalments, this year’s workshop facilitated a span of public computer vision challenges to bring new innovation to assistive technologies. The workshop also included a series of panel discussions with notable computer vision researchers, access technology advocates who are blind, and industry specialists to discuss the past, present and future of these technologies. The Visual Question Answering (VQA) Challenge, the workshop’ founding challenge, invited teams to develop a model that could automatically answer a question about an image which a person who is blind/low-vision has captured on their mobile phone. This year, 44 teams participated in the challenge, with the top-two winning submissions belonging to the teams from 24 DAILY CVPR Workshop Wednesday
Made with FlippingBook
RkJQdWJsaXNoZXIy NTc3NzU=