with
We have been recruited by Huddly for this project. Huddly's goal is to enhance the experience of video communication by bringing intelligent cameras into the meeting room. They do this by carefully combining hardware, software and artificial intelligence solutions to produce conference cameras with intelligent features.
Contrary to regular cameras, these can not only provide a video feed, but by utilizing an onboard neural engine is able to see, understand, process and respond to what they are seeing. This means it can automate tasks like controlling the cameras itself, provide high-quality analytic data about what it sees and allow for computer vision models to be deployed directly on the camera.
what
Thanks in part to the onboard neural engine, the cameras capabilities are virtually endless, and thus Huddly are always on the lookout for new intelligent features to implement on their cameras. This is where we come in.
We want to look into the possibilities of using exclusively hand gestures as a means of interacting with a computer. By moving user functions away from traditional means such as keyboard and mouse, to relying simply on gesticulation, one could provide a profusion of ease of use and flexibility to the end user. This is especially true for applications in video communication, in which a video feed already would be present.
We will therefore spend this semester making models and researching the possibilities and limitations with neural network based gesture recognition, mainly focused on it's applications in video communication.