Your daily frequent11y newsletter, brought to you by @ChrisBAshton:

Developing Real-Time, Automatic Sign Language Detection for Video Conferencing

• Google intern Amit Moryossef writes about a research paper on real-time sign language detection using human pose estimation. The model, called PoseNet, reduces input from a HD image to a small set of landmarks on the user’s body, which can be analysed frame-to-frame to predict whether a user is signing, to an accuracy of 83-92% – you can try this for yourself in the demo or watch the video.
• As conferencing software tends to focus on someone who is speaking verbally, it is hoped sign detection could be used to move focus to someone communicating via sign language. The researchers found a clever way of triggering the focus: when the model determines that a user is signing, it passes an ultrasonic audio tone. The audio is transmitted at 20kHz, which is outside the hearing range for humans, but fools any video conferencing application into thinking the user is speaking. This research could be used to better include those who communicate via sign language.

Prefer longer newsletters? You can subscribe to week11y, fortnight11y or even month11y updates! Every newsletter gets the same content; it is your choice to have short, regular emails or longer, less frequent ones. Curated with ♥ by developer @ChrisBAshton.