Crowdsourcing images of hands, GiveAHand.ai aims to bridge the AI gap for sign language

While the AI landscape is developing at breakneck speed, capabilities surrounding sign language recognition have been hindered by a lack of well-structured and extensive data sources. To fix that problem, the American Society for Deaf Children has launched GiveAHand.ai, an ambitious project to create the world's largest open-source image library of hands, complete with tagged data.

Anyone can upload an image of their hand using the website and a webcam. Participants hold their hand in whichever gesture they choose, and a picture is taken and uploaded to the database. Next, the image is tagged, and keypoints are added to define the location and shape of each finger. Researchers and developers can download the full, structured data set to create more accurate hand and finger detection models.

Since the goal is to democratize access, GiveAHand.ai's database will be available for both commercial and non-commercial use. The initiative was developed in partnership with creative studio Hello Monday (part of DEPT®), which previously worked with ASDC on fingerspelling.xyz.